Dr. Suit: Good morning, everyone. This is Herman Suit and it is January 18th 2013. I'm representing ASTRO and, together with Marcia Urie and Lynn Verhey, we are interviewing Michael Goitein. So, Michael, to start out, tell us a bit about your early upbringing and schooling. Were you set on a path which led you directly to your eventual career?
Dr. Goitein: Well, yes and no. When my elder daughter, Lara, was contemplating what subject to major in when she went to college, I asked a number of people I knew whether they ended up in the career that they initially imagined they would pursue – and almost all of them reported that they did not. Their eventual careers owed far more to serendipity than to forethought and planning. I certainly did not plan to be a medical physicist. I didn’t even know the profession existed. But, on the other hand, from an early age I was interested in technology as demonstrated by my pulling apart various devices to inspect their innards. One of these I remember was a treasured family clock - which never subsequently worked.
Neither of my parents was technically minded. My father was a law professor at Birmingham University and was quite impractical. I do not recall ever seeing him with a hammer or screwdriver in hand. My mother was multi-talented: an artist; an interior designer and amateur (but very good) architect; and a very practical (though not terribly successful) business woman. I was, however, nudged in the direction of pursuing the sciences by two factors. First, from an early age I showed an aptitude for mathematics. Second, the English school system tended to stream children into either the arts or the sciences at a fairly young age, and my choice was made at an even younger age than usual due to the fact that, as a bright youngster, I had been advanced to classes two years ahead of my peers. The choice of sending me down the science path was, as I recall, made with little input from me. I’d like to interpolate that I think advancing a child in such a manner is a mistake and is one that I feel I paid a price for during much of my life.
At the end of my school years I obtained a scholarship to go to Balliol College in Oxford University. I decided, pretty much on my own initiative as I recall, that I was too young to go up to Oxford directly and that I should spend a year growing up. (Actually, a year was nowhere near long enough. Bertrand Russell famously said that “Education is wasted on the young.”) I decided to spend a year at the Sorbonne in Paris, studying mathematics. I had hoped that this would put me in proximity with French students (which it did) so that I would come to socialize with them (which I did not). But, I did learn a little mathematics. To this day there are mathematical concepts and terms whose names I know in French but not in English.
Dr. Suit: That takes us up to Oxford. What did you study there, and did you socialize with those students?
Dr. Goitein: Well, Herman, I think you know from your own experience at Oxford that you have asked the questions in the wrong order. I quickly deduced that a primary goal of an Oxford education, at least in those days (1958), is socialization. That is, in the case of boys, to turn one into a young gentleman. So, yes, I socialized: I learned to keep a sherry decanter filled with a fairly decent sherry always close to hand in my rooms; I attended the Oxford Union’s always interesting and often frivolous debates; I went regularly to the Jazz Club’s weekly bash; and so forth. I also went to a lot of fascinating lectures in subjects far removed from my major. I particularly remember Isaiah Berlin’s inaugural lecture entitled “The two liberties” and a series of talks on the 18th century English artist William Hogarth, which were so popular that they had to be held in the local cinema in order to accommodate the large number of listeners.
To answer the first part of your question, I elected to study physics. In some ways this decision was made a bit difficult for me. I had been accepted to Balliol to read either physics or chemistry. Thinking that they may have formed an impression of my abilities, I asked for separate interviews with the physics and chemistry senior tutors at the college, primarily to ask them which specialty they thought I should pursue. Of course they did not come right out with a clear suggestion but, on the whole, I got the impression that the physics tutor favored chemistry and the chemistry tutor favored physics for me. So, I was thrown back on my own resources. In the end I chose physics on the grounds that it was the more fundamental discipline. I’ve never regretted that decision, although modern chemistry has advanced way beyond the horizons that I sensed for it in those days. For one thing, they have learned how to build the most amazing, highly specialized, molecules.
One’s degree at Oxford depended entirely on one’s performance on the final exams at the end of three years. The distractions of the place meant that one’s learning was crammed into reading during the fairly long holidays and some serious cramming in the last few months. Degrees were classified as first, second (split into two parts) and third class honors. I was awarded a so-called “good second” for my performance. My tutor was surprised at this, having expected me to get a first, and I too was disappointed. However, I have come to consider this to have been a correct grade. I have come in contact with, and in a few cases known well, some of the best minds in physics of my time – such people as Ramsey, Purcell, Schwinger, Glauber, Gilbert, Alvarez, to mention some of them and I know my talents are in no way in their league, nor in the next league down. I am comfortable with this realization.
Dr. Suit: Well, we’ve got you through Oxford. What then? Did you feel you needed another year’s break?
Dr. Goitein: Herman, you have hit the nail on the head. I had arranged to do graduate work in low temperature physics with Kurti at Oxford. However, I felt that, after the ardors of an Oxford undergraduate education, I first needed a year off and I was lucky enough to get a fellowship for a year’s study of physics at Harvard. My plan was less to study physics than to experience life in the United States. With my Oxford arrogance I ‘knew’ that American students were far less well prepared than were we British, so I anticipated that the class work would be a breeze and would in no way hamper my real goal. I was in for a huge surprise. The teaching was excellent, giving me contact with some of the finest possible minds; the students were bright and knowledgeable; and, more than anything else, for the first time I found the subject really interesting. I abandoned my plan to experience America and settled down to learn some physics. I also abandoned Kurti and Oxford and decided to remain at Harvard for the rest of my graduate education.
I chose to work in elementary particle physics, the discipline dedicated to understanding the nature of sub-atomic matter. I didn’t imagine that I would be the one to uncover the secrets of matter, but I thought that I would like to be able to understand them when they were discovered. Alas, this was an unrealistic expectation; the theories today are way beyond my ability to comprehend. In any event, The Cambridge Electron Accelerator, a 6 BeV electron accelerator, had just been built on Harvard’s grounds, and I joined a group doing experiments designed to explore the charge structure of protons and neutrons. This was done by scattering high energy electrons from the accelerator off liquid hydrogen or deuterium and measuring the angular and energy distribution of the scattered electrons. These were wonderful times. Together with a fellow graduate student and good friend, Bob Budnitz, I learned a lot, worked hard, and had fun. I was a computer freak and, amongst other things, was responsible for the online data collection and computer analysis. Eventually, I wrote my thesis using a computer word-processing program that I had concocted and submitted Harvard’s first doctoral thesis composed on a computer. It was also, I believe, the longest thesis that the physics department ever had to deal with, and one of my three examiners, coming across me in the hallway a few days before my oral exam, said to me “Michael, do you really expect me to read all this?” I hastily put together a synopsis for him.
Dr. Suit: When you look back on your days as a graduate student and on your thesis work, how do you view them?
Dr. Goitein: As I’ve hinted at, I had a very broad experience of many different technologies, and I learned a lot from that. But I am a bit doubtful about the importance of what we learned. Today, some 40 years later, our work is largely forgotten or ignored and I don’t think it made an appreciable impact on the current understanding of nuclear structure. I have learned that this is not all that surprising. The essence of scientific research is that one does not know how it will come out. Much of it turns out, unavoidably, to have been unproductive. Research is very inefficient; only a handful of experiments or theories turn out to change our ideas or introduce new possibilities. This makes it quite frustrating at times, but success is all the more precious because it is so rare.
Dr. Suit: And, after you graduated?
Dr. Goitein: I spent one-and-a-half years as a post-doc at Harvard, participating in a very enjoyable experiment led by Gene Engels at Brookhaven National Laboratory. I then joined the staff of what is now called the Lawrence Berkeley Laboratory (LBL) in Berkeley, California. I was seconded to the Moyer-Helmholz group and worked on an experiment to test time-reversal invariance. The laboratory is situated in a wonderful setting half-way up the Berkeley hills with a view over the San Francisco Bay which, particularly at sunset, can be stunningly beautiful.
But, despite the great scenery, I began to grow disenchanted with the work I was doing and with what I saw to be the direction in which the field of elementary particle physics was going. Experimental groups were getting to be bigger and bigger – by then they were typically from ten to 20 strong. Today there are reports of experiments which list quite literally many thousands of co-authors. I began to feel increasingly inessential and replaceable; and that my work lacked social value.
Dr. Suit: How did you react to these realizations?
Dr. Goitein: I decided to change fields. I had had a little insight into the general field of medicine through the fact that my wife of that time was a physician and I had a colleague, Howard Maccabee, who had trained as a medical physicist with Gordon Brownell at the Massachusetts General Hospital (MGH). More importantly, LBL had a strong biophysics department led by a very inspirational man, Cornelius Tobias. So, I went to talk with him to see if there was a chance of joining his group. Being a wise man, he simply mentioned a few problems which interested him. One of these was the then unsolved problem of computed tomography (CT). The problem was whether, and if so how, one could reconstruct the interior structure of an object (e.g. a patient’s anatomy) from a series of projections (e.g. radiographs) taken from multiple viewpoints around the object. Tobias was so sure it should be possible that he had one of the physicists in his group, John Lyman, put together a phantom and make some projection measurements of it – but they had no idea as to how to analyze the measurements. I thought about their problem while seated in Toby’s office and thought that I knew how to solve it. And, I made the mistake of telling him so. Later, back in my office, I realized that I had made an error in my mental calculation and my ‘solution’ would not work. So, now I was in trouble. I mulled the problem over for a few days and came up with an alternative approach which did, indeed, work. I performed a number of computer simulations, analyzed Lyman’s data and brought the results back to Tobias. He was happy with my results – but, he never offered me a job! I published my method and became, as I later learned, somewhere between the sixth and tenth person to have solved the problem of CT – for which the first two inventors, Hounsfield and Cormack, got the Nobel prize.
Later, I collaborated with Doug Boyd at Stanford who had developed a fan-beam CT scanner, but did not know how to analyze the data – previous devices used a laterally scanned pencil beam yielding a parallel geometry for the measurements. I was easily able to adapt my method to the fan-beam geometry and provided Doug’s machine with an accurate reconstruction – a process for which we obtained through Stanford a patent. Of course, fan beams now provide the normal geometry for CT.
Let me interject here, although it’s a bit out of place, a few words about a study of the value of CT in radiotherapy that I later undertook. Health care payers were, in the early days of CT, extremely concerned about its cost. A machine went for about a million dollars and, in those days (the 1970’s), that was an unimaginable amount to pay for a single diagnostic device. So they put roadblocks in the way of purchasing them. The MGH acquired one of the early whole-body scanners, and part of the justification was that it would be used as a research tool to evaluate CT’s clinical value. I was therefore able to get time on the machine and, with Jack Wittenberg, a radiologist, and Marta Mendiondo, a medical physicist in our group, we designed and conducted a study in which patients, already imaged and planned using conventional techniques, then received a CT study which was used to evaluate the adequacy of the original plan. This was the first prospective study of the impact of CT on radiation treatments and we found that the pre-CT plans were inadequate in 52 percent of the cases. I then performed a modelling exercise, reported in a paper in JAMA, estimating the impact that the use of CT would have on outcome. The model predicted a 6 percent improvement in local tumor control. Today, of course, there’s hardly a cancer patient who does not have not just one but several CT studies in his or her folder. I like to think that our study has some modest impact on the acceptance of CT into clinical practice.
Dr. Suit: Since Tobias did not bite, what did you do then?
Dr. Goitein: My entry into medical physics was not without some hiccups. Tobias had advised me that, if I wanted to contribute to cancer care, I should involve myself with mice rather than men. Patient treatments are too messy, he opined; far better to work in a more controlled way with animals. However, I felt that, if I wanted to contribute to patient care, I would rather get as close to the patient as possible. So I decided to try to get a job as a hospital physicist.
I wrote letters to, as I recall, 24 institutions which had medical physics programs of some sort. Most did not even answer my letters; the rest sent regrets. I asked Howard Maccabee for suggestions and, one day, he came and told me that he had heard that Ted Webster at the MGH might be looking for someone. I phoned Ted immediately and let him know of my availability. “There’s a high energy physicist on every street corner these days.” he said to me, “Why should I choose you?” Well, this was an inauspicious start but somehow I managed to convince him to take me on. Then, out of the blue, the new chief of radiology took the post away from Ted, needing it for another purpose. Ted rang me up, terribly apologetic, to un-invite me. He told me, though, that a new young chief of radiation therapy had just been appointed and he might be interested in me; he’d pass my name along. That, of course, was you, Herman. You invited me to an interview. I brought along some computer simulations that I had made to show my interest, suggesting a possible advantage to treating with multiple radiation beams entering the patient near-isotropically and so spreading the extra-tumoral dose around over a greater volume, but at a reduced level as compared with beams confined to a single plane. In retrospect, this was a bit paradoxical in view of my current concern about the delivery of dose baths to a large volumes of a patient. In any event, it did not hurt that you are an Anglophile and you offered me a job in spite of my total lack of training or experience in medical physics. You advised me to read Johns’ and Cunningham’s medical physics textbook before I came, which I did from cover to cover, and armed with that I showed up at your door in early 1972 and stayed with you for just one month short of 30 years.
Dr. Suit: While we’re on the subject of your lack of training in Medical Physics, I think it might be kind of interesting if you, Michael, would comment on your and Ed Epp's policy that there would be no graduate student PhD thesis work in our department. That is, that everyone coming to a staff position would already have a PhD in pure physics. Now this is something that I thought was a fine idea and it seemed to work well. But, things have changed a lot and I want to know if you would comment on your present view.
Dr. Goitein: I did feel - and it was in view of what my own education had been that one got a more in-depth training when one followed one of the more usual physics routes. I thought that our research program at the MGH did not have the rigor and depth that I had experienced in my elementary particle physics training. It seemed to me to be unfair to allow people to think they were getting a doctoral-level training when, in fact, we couldn’t really offer that. On the other hand, George Chen changed the policy and started bringing in graduate students and I think he made it a very lively place for them. So I must admit that in later years I had some second thoughts. I think in the end it has a lot to do with personality. You know, if a person feels that they can offer a proper environment for a PhD training, then I think they should go ahead. I was a bit surprised that Ed Epp agreed with my viewpoint because he himself had a very substantial radiobiology research program.
Dr. Urie: I'd like to ask a follow up to that. What is Michael's view now on the requirements for residency, medical physics residency and having to go through a medical physics graduate program to be board certified?
Dr. Goitein: Well, I'm blissfully ignorant of exactly how things are today, so I can't quite answer. In the generation before me, people like Ted Webster had no training at all in medical physics. They came in having trained, in his case I think, in nuclear physics and they picked up medical physics on the way. This is more or less what I did. It didn't work out so badly as a way in because medical physics in the trenches strikes me as something of a craft, if you will. There's a lot to know; I'm not denigrating in any way the amount of knowledge that one has to have. But I think it might be better to split the problem into two parts: a basic training in ‘pure’ physics at either the masters or doctoral level depending on the level the person wishes to reach, followed by a period of practical clinical training.
Dr. Verhey: Thank you, Michael. For what it's worth, I've always given you credit for the fact that I had exactly the same point of view at UCSF. We never took a physics resident in who didn't have a PhD in a more pure field than applied medical physics. Most of my colleagues or people of my age disagreed with that, but I think it was a good idea and I got it from you.
Dr. Suit: Okay. Well, that's a very nice clarification. To change the subject, what were your impressions and experiences when you first came to the MGH?
Dr. Goitein: Of course everything was brand new to me. The department was bustling with activity. I was extremely lucky that the two medical physicists already on board, Miriam Gitterman and Art Boyer, gave me a warm welcome and patiently initiated me into the myriad details of the work. The department was cramped for space. A new cancer center was being built to house it, but for the present it was mainly located in the few rooms coming off a rather short corridor, with scarcely space to swing a cat, let alone a newly arrived medical physicist. I was given a corner of the control area of one of the three treatment machines, an orthovoltage (280 KeV) machine. It was literally in the corner; my desk was a small triangular piece of Formica put in by the hospital’s carpenters in a corner where two walls intersected. What was so fortunate was that I could observe indeed, could not avoid observing the treatment process up close. I was within arm’s reach of the technician, saw the patients as they entered and left the treatment room, observed them through a leaded glass window during the time that they were positioned and treated, and heard everything that was going on. There could not have been a better learning experience. I noted how useful it was that the treatment head was articulated so that it could be angled to bring the beam in from any direction not just within a plane as the more modern isocentric machines tended to be limited to. It was there that I saw my first error in treatment delivery: a patient was treated using the technique intended for a different patient. An easy mistake to make and this led me into the field of computer monitoring of treatments (so-called ‘Record and Verify’ systems) which had the goal of avoiding such mistakes.
I will never forget, Herman, your coming to chat with me in my little corner after I had been at the MGH for a few months. You asked me what I had been doing and then, clearly unimpressed with my accomplishments to date, urged me to take on some research projects and build up a record of professional activity. It was as though you had put me into gear, and I don’t think I ever got back into neutral for the entire time that I was at the MGH.
Dr. Suit: So, tell us what your early projects were.
Dr. Goitein: If I may, let me first step back and make some general comments. I have been thinking about what my overall approach has been, and I think that it has informed most of what I did. There were four main strands:
- At first unconsciously, and later consciously, I decided not to attempt to tackle the prominent problems of the day. There are always ‘fashionable’ problems and people tend to gravitate towards them. Since so many people are already working on them, one can be pretty sure that others will solve them and there’s no sense in simply racing to be the first to do so. For example, accurately computing the ‘dose at point P’ was the holy grail when I entered medical physics – so I elected to focus primarily on target and organ coverage rather than dosimetry near the center of the irradiated volume.
- I appreciated that medical physics is an applied field – and that therefore the application is king. Therefore I always asked myself whether the solution of any problem that I was contemplating working on was likely to improve cancer care. If not, or even if its probable impact was simply doubtful, I backed away and conserved my energies for something more relevant to the goal I had set myself. Having said which, one has to be careful to keep one’s expectations at a reasonable level. It is not given to everybody to find the cure for cancer. I told myself that I would be content if, over a lifetime’s career, I would have improved survival of our patients by an average of two percent. If fifty people achieve two percent, then the problem’s solved!
- I tried to maintain a broad focus. Although I became known for working with protons, I consider my professional work to have been at least as relevant to conventional (X-ray and electron) therapy as to proton therapy.
- I have consistently sought out collaborators with whom to work. You three are foremost among these and it has been a huge pleasure to have collaborated with you. At the time I was awarded ASTRO’s gold medal, I thought I might show the names of all those with whom I had actively collaborated in order to thank them, so I started to compile a list. I stopped when it reached 231 names. I realized that there certainly were others whose names I had not yet recalled and I abandoned the idea of putting up a slide. And, too, I hate the NIH (not invented here) phenomenon. If others have already solved a problem, or are well on the way to solving it, then one should take advantage of their efforts and use one’s own energies elsewhere to better advantage.
In this connection, I once heard the saying that “the inventor of something is the first person who forgot to say from whom he learned it.” Ever since hearing and digesting this saying I have tried to give attributions where they are due. Many, perhaps most, of my ‘innovations’ were not wholly original. Others had the same or similar ideas (of which I was not always aware). My contribution was often to bring these ideas into general clinical practice, and with (relatively) easy-to-use tools.Dr. Suit: Back to specifics. What projects did you undertake initially?
Dr. Goitein: Let me mention just two things:
When I first came to the MGH I noticed that the treatment couches were covered with several-inch-thick foam mattresses. While these were comfortable for the patient, they did not provide reproducible positioning; the patient tended to lie in a slightly different position at each treatment session. I developed a much thinner, though still reasonably comfortable, mattress with a non-slip and washable surface which led to more reproducible positioning, and this was widely adopted by industry.
Reproducibility of the position and orientation of a patient’s head was also imperfect; the tools available being various gadgets, augmented by lots of adhesive paper tape. With the help of Julian Cherubini, a plastics salesman, I developed a mask which was made of thermoplastic sheets which could be easily formed to fit the individual patient and which were perforated by multiple holes, both for comfort and to partially protect the patient’s skin from the effects of radiation. It was gripped in a wooden frame (initially I used a Wilson tennis-racket frame) secured to the couch top by quick-release fasteners. This invention has also been widely adopted and these masks can now be seen in virtually every modern radiotherapy department.
Of all my developments, the mattresses and masks, simple as they are, rank amongst the things I have done about which I am most pleased.Dr. Suit: One of your interests was in developing a 3-D treatment planning program. Can you tell us about that?
Dr. Goitein: Of course, but I have one related subject that I would like to address first, namely the issue of uncertainty. It does have a lot to do with treatment planning in my view.
In my opinion, no physicist worth his or her salt, and indeed no scientist, should state the value of a measurement or calculation without adding an estimate of the uncertainty in that value at a specified level of confidence. When someone states that they know what the dose will be, or was, to a given part of some structure, let us say 48 Gy, they are probably reporting a value taken from some pretty picture generated by their treatment planning computer. Let us call this the ‘nominal dose’. Assume for simplicity that both positional and dosimetric errors are normally distributed. The nominal dose will be our best estimate of the actual dose. However, under these assumptions, saying that the (nominal) dose is 48 Gy is equivalent to saying that “There is a 50 percent chance that the dose will be greater than 48 Gy.” Given the latter statement, most clinicians would gulp and immediately ask “Well then, how much greater than 48 Gy can it be?” In effect they would be asking for (and should already have been given) the confidence bounds on the statement of dose.
I recall walking through the treatment area one day and seeing out of the corner of my eye a couple of residents conferring on a case. Somehow, I realized that their decision that the port film indicated that the patient was correctly aligned was erroneous. This led me to design a study, performed by two South African medical students who were visiting our department for a few months, in which treatment portal films taken over the course of treatment were compared with one another and with the initial simulation film. This study documented the degree of patient misalignment, both day-to-day and relative to the intended alignment. It was, I believe, the first study to identify that systematic errors can be, and often are, larger than random errors in patient positioning – and hence more worrying.
In my development of a treatment planning program I included an estimate of the dose uncertainty (usually at the 1.5 standard deviation level). Treatment devices such as apertures and compensating boluses could be designed with safety margins based on the computed uncertainty estimate. The program included methods of displaying uncertainties in the dose distributions. I also advocated that normal tissue constraints be stated and imposed in terms of the estimate of the upper bound of dose and not on the nominal dose. Alas, it is relevant to report that my clinical colleagues eventually gave up this way of imposing and reporting dose constraints. Their reasoning was that they became weary of trying to explain this unusual approach to their bewildered colleagues.
It has been a great disappointment to me that these or similar methods of uncertainty analysis have largely been ignored by practicing medical physicists and the supporting industry.Dr. Suit: Speaking of uncertainty, I would like to just comment, Michael, that it seems to me that it would not be out of line for you to put a little bit of emphasis on the fact that the first patient we treated by proton beams had position confirmation about as accurate as I think anything going today for lesions in non-mobile sites. I was at a DoE-NCI conference last week and a very prominent radiation oncologist said that, when protons started at Boston, they had very little knowledge of the position of the beam relative to the target. I responded with a little bit of vigor saying that I think it would be difficult for anybody to do much better today than we did then because we had biplanar imaging before each field, although that required reading of the film followed often by re-positioning of the patients and repeating of the filming as indicated to achieve the planned position accuracy. Further, a large fraction of the patients had fiducial markers placed as well. Additionally for the ocular melanoma patients, Evangelos Gragoudas, our ophthalmologist partner, sutured usually four quite small tantalum rings around the tumor base for positioning by X-ray beams while the patient gazed at a very small and precisely positioned light beam. Then during the proton beam-on time, a greatly enlarged image of the iris was viewed continuously; if there was any significant eye motion the beam was turned off immediately. The local control result of 95 percent at 15 years is convincing evidence that the beam was on the target. This procedure was largely developed by you, Michael.
Dr. Goitein: Well, Herman, I think you have made the point perfectly. We used radiographic alignment on almost every patient – except for a very small number of patients with quite superficial lesions where we went by skin marks. The reason for that is because of two things: 1) the whole rationale of proton therapy is to confine the dose as well as you can to the volume that you want to treat and have as little as possible outside, so the more accuracy you have the better you can achieve that goal; 2) when the beam is mis-registered with the patient, then the compensation is compromised. Lynn, here, spearheaded a very nice study looking at radiographs taken before and after treatment to determine by how much patient alignment could change and the numbers were in the millimeter range. They varied a bit on the method of immobilization and on whether the patient was seated or supine. We did realize excellent alignment and I think it was the right thing to do.Dr. Suit: Well, to return to treatment planning, how was it that you decided to develop your own treatment planning program?
Dr. Goitein: When I arrived at MGH, treatment plans were designed using a teletypewriter connected to a remote computer at Memorial in New York. The system was developed by Radhe Mohan and was the mainstay of many institutions for many years, but by today’s standards it was very limited. Art and I had the job of selecting the department’s first in-house treatment planning computer. We chose an interactive system made by a small firm, Artronix. It was made interactive though the manipulation of 4 knobs which controlled, as I recall, gantry angle and the collimator settings. Patient anatomy was defined using a digitizing arm, but only in a single plane. The skin surface was traced from the shape of a solder-wire which had been bent around the patient and internal anatomy was outlined by some artful combination of information from radiographs and from an atlas of human anatomy based on sectioned frozen cadavers.
Although considered state-of-the-art at the time, the Artronix system was quite primitive. It could not directly use imaging information such as CT images which were becoming available in the early 1980s; it was at best what I called 2½ dimensional only supporting treatment techniques in which all beams lay in a plane; and it only allowed one to compute dose distributions in a single, or in a very few, transverse sections. It was clear that a much more capable system was possible, so I decided to try to develop one.
This decision was catalyzed by my experience in planning the treatment in 1974 of the second patient to receive proton therapy at the Harvard Cyclotron Laboratory. She was a lady with a chondrosarcoma of the skull base which required the use of a posterior-oblique beam. I spent three days and much of three nights designing her plan (based on a sequence of so-called smearing tomograms) and yet more time implementing it. At the end of this effort I pledged to myself that I would never again do by hand what could be done by computer. I learned from this experience that invention must often be driven by inherent laziness. Inventors seek solutions which will allow them to accomplish a desired goal with ease.
The development of my treatment planning system spanned a decade or more during which several important technical developments took place: the development of powerful and, eventually, affordable mini-computers (specifically Digital Equipment’s VAX computer with its virtual memory, costing when I bought it about $250,000 – today a comparable computer system could be bought for about $1,000); the development of interactive computer graphics (an outgrowth of the space program with its military underpinning); and, of huge importance, the invention, development and proliferation of computer tomographic scanners (CT) with their amazing anatomic detail. In connection with CT, I went to a conference in Puerto Rico in, I think, 1973 or ’74 at which whole-body CT images were shown publicly for the first time. I was so impressed by them that I managed to snag several CT images (of Mike Terpegosian) and carried them around in my briefcase for over a year, showing them to everyone I came across. I knew right away that CT would revolutionize radiation therapy.
Dr. Suit: Your treatment planning program introduced many new ideas. Can you list for us some of its main innovations?
Dr. Goitein: I’d be happy to. But first, let me make a general point which informed all my developments and is not something that, even today, all software developers recognize. Namely: (1) Even if you have a great idea of what you want to accomplish, if you don’t have the necessary tools, you won’t be able to do it. for example, if non-coplanar beams are not supported, you will not be able to design non-coplanar treatments; and (2) even if one has the tools, unless the advantage is truly enormous they won’t be used unless they are easy to use .
Having said which, and bearing in mind my point about how much I owed to previous innovators and to those who worked with me. Mark Abrams, Greg Moulton and Tom Miller deserve special mention. I’ll rattle off brief descriptions of some of our innovations. Our planning program included:
- fully three-dimensional geometry (patient and delivery equipment) this, for example, made the planning of image-based non-coplanar treatments possible;
- use of CT images, based on which the three-dimensional shape, size and location of the tumor and of the normal anatomy could be defined;
- a soft-labeled user interface (a video screen surrounded by push-pull switches whose function was labeled by the computer);
- interactive tools to delineate anatomy from the CT data;
- modeling of the treatment equipment which allowed the user to move it interactively (a capability later termed ‘virtual simulation’);
- display of anatomy in the beam’s-eye view (prior work had been done by Stirling, Siddon, and McShan);
- interactive and automatic design of ancillary devices (e.g. apertures, compensating boluses) in the beam’s-eye view;
- back-projection of aperture and collimator outlines on sectional CT images (transverse, sagittal and coronal);
- dose calculated three-dimensionally;
- dose displayed as either isodose contours or, preferably, as color-wash on transverse, sagittal and coronal images;
- computation and display of dose uncertainty bounds;
- treatments designed taking uncertainties into quantitative account;
- what came to be called digitally reconstructed radiographs (similar views but without perspective distortion had been developed for diagnostic use);
- dose-volume histograms (invented together with Lynn Verhey);
- plan comparison techniques, e.g. side-by-side display of two or more plans (in color-wash or with isodose lines); overlay of DVH’s for two or more plans for a particular volume of interest; and side-by-side comparison of biophysical quantities (TCP, NTCPs);
- implementation of the capability to interact with treatment plans during departmental conferences and from a remote site, namely from the Harvard Cyclotron Laboratory.
The routine clinical use of all these capabilities was, I think, novel. Most of them are now in day-to-day use worldwide – but, alas, not uncertainty analysis.Dr. Suit: Didn’t you eventually collaborate with a company in the development of a commercial treatment planning system. What happened with that?
Dr. Goitein: Being impatient to see my ideas in clinical practice, I, together with Peter Kijewski of Harvard’s Joint Center, formed a collaboration with Siemens to develop a commercial treatment planning system. It was to be based on my system and on Peter’s ideas for, primarily, data management.
This effort lasted at least five exhausting years and finally failed. The project was terminated by Siemens for two main reasons: (1) due to problems that Siemens were having in another of their plants which made ultrasound scanners, they were having trouble getting FDA clearance for their projects in general; and (2) our project was taking too long to come to fruition. In part, I blame myself for the second of these reasons. I simply tried to do much in the first release, fearing that, if I didn’t get all my ideas in the initial system, then they would never get in.
Although our development was discarded, it did, I think, have some influence. The beta version was shown at several congresses and was seen by large numbers of viewers, commercial and otherwise. And a few beta versions, I think four, were distributed, one of them to the University of Zurich’s radiotherapy department and one in Munich, and were generally very well received. I think it is no accident that many of our developments showed up in subsequent commercial planning programs.
Dr. Suit: There was another collaboration you were involved in, namely in some NCI-sponsored working groups. Tell us about that.
Dr. Goitein: Yes, that was an interesting interlude. It took place during the 1980s and was, I think, quite important for the dissemination of ideas about three-dimensional treatment planning. There were two projects, each involving collaborations of four institutions; the MGH participated in both. The first was initiated by David Pistenma during the time that he was head of the radiation development branch of the NCI. He hoped that one could decide on the relative merits of various treatment modalities by a paper computer exercise – and it would be faster and cheaper than funding a lot of clinical trials. So he let a series of contracts for a working group to engage in the inter-comparison via treatment planning of X-rays, protons, Carbon ions, neutrons and, initially, pi-mesons. However, one did not then, and still does not today, have the basis for assessing the clinical importance of the radiobiological differences between the particles. Consequently, the collaboration, although valuable for stimulating interest in 3-D treatment planning, did not succeed in its primary goal.
The project culminated in a voluminous report, published by the NCI, and several hundred copies were disseminated. That the interactions between the groups were very fruitful can be deduced from the titles of the technical sections of that report which I think one can say were, for their time, groundbreaking. The titles were:
- Description of evaluations for several sites
- The use of CT and other modalities in planning treatment
- Correction for inhomogeneities
- Radiobiological considerations and microdosimetry
- Uncertainty analysis
- Normal tissue tolerance
- Tumor control probability
- Evaluation methodologies
- Tape exchange standard
As regards publication of the findings in the open literature, I have to make an embarrassing confession. I undertook to put a manuscript together, which I did, and I submitted it to the International Journal of Radiation Oncology, Biology, Physics. It came back with a multitude of reviewers’ complaints. Being tired of the project, I postponed dealing with them and, eventually, the manuscript died in my desk drawer and was never published. To this day I feel guilty about this failure of mine which was very unfair to the other participants.
The second project was developed by Al Smith who had taken over by then from Dave Pistenma. The collaborators were to define the elements of three-dimensional treatment planning in the context of X-ray therapy. Of course, the first project had gone a long way towards that end, but there is no doubt that the second project made significant extensions to that work. For me, however, the dominant memory of the second project was an unfortunate one: a lengthy squabble regarding the publication of results. This took place at the start of the project and very nearly put an end to it before it had really begun. Most of the participants wanted to be sure that all four institutions would have a mathematically equal presence of their members in the list of authors and would be the leading institution in one quarter of the publications of which, in the end, there were 20 (a number divisible by four you will note!). They wanted the author lists to be decided from the beginning. I resisted this notion strenuously, believing that authorship should reflect actual substantive contribution. In the end, in order to not bring the project to its knees, I caved in. But, to this day, I have a bad taste in my mouth. It seemed, and still seems to me, to be the antithesis of what an academic collaboration should involve.
The MGH was nearly thrown out of the project for another reason. There was a lot of debate at the beginning about the dose calculation algorithms. The other groups had implemented complex algorithms and believed my simpler one, as they judged it to be, was inadequate. They felt that the MGH’s planning program therefore did not meet the minimum required standard. They did agree, however, to delay our expulsion until the results of comparisons between calculation and measurement in a variety of phantoms had been ascertained. What they did not understand was that my algorithm took great care of defining how dose falls off at the edge of a field. They were focused on ‘dose at point P’ issues where P tends to be near the middle of the field. It gave me no little satisfaction when the MGH algorithm outperformed theirs in the majority of the tests.
There is no doubt in my mind that the two projects, taken together, were very influential in introducing and further developing three-dimensional treatment planning. The second project, in particular, mapped out the field of 3-D treatment planning. It was both a valuable status report and provided a template for future developments. Its results filled a full issue of the International Journal of Radiation Oncology, Biology, Physics (volume 21). Until then, 3-D treatment planning was in its infancy. No 3-D system was commercially available, and very few people even knew what it involved. And, the collaborations between the groups led to cross-fertilizations which affected their ongoing developments in important ways.
Dr. Suit: You have mentioned biological modelling. I know you had a great interest in this area. How did you get into it and what were your contributions?
Dr. Goitein: In my later years, I coined a phrase along the lines of “Dose is just a surrogate for what we really care about, namely the extent of biological damage to the tumor and to normal tissues.” I have seen and heard the word ‘surrogate’ used often enough to have the feeling that I played some role in putting this concept on the map. Of course it was always obvious to the more sophisticated radiation oncologists, but the rank and file of physicists and physicians seemed unaware of it.
I appreciated this point the moment I entered the field. Initially I focused on the issue of fractionation. Ellis’s Nominal Standard Dose concept was popular at the time, and a couple of derivative formulations, TDF and CRE, had evolved from it. These dealt, however, with regularly-spaced fractionation with equal doses per fraction. I wondered how to deal with irregular spacing and unequal doses per fraction, as not infrequently occur in the clinic, and I developed and published an approach to this problem. However, a conference in Wisconsin devoted to time-dose fractionation emphasized for me the complexity of the problem and persuaded me not to pursue work in this area.
Some years later, though, I came back to biophysical modelling. It came about through the wise words of Sten Graffman, a Swedish radiation oncologist who was spending some time at the MGH (and who, sadly, just recently died). I had been worrying about the dose perturbations which our studies had taught us that protons suffer when passing through inhomogeneities. It’s quite possible that the dose in a small volume - in the shadow, for example, of a thin piece of bone – could suffer a reduction of 20 percent or so. In those days the conventional wisdom was that tumor control was driven by the minimum received by a tumor; any dose above the minimum to other parts of a tumor was, it was supposed, ‘wasted’. If this was true then, I worried, if a reduced dose happened to lie within the tumor, the tumor control probability (TCP) could be badly compromised. Upon my speaking of my concern to Sten, he answered, “Well then, why don’t you try calculating by how much tumor control would be reduced?” It turned out that he had previously done some work on calculating tumor control probabilities and that got me started.
I don’t want to go into details of the model I developed, but it had some characteristics that I should mention. In contrast to the few models in the literature which dealt with uniform irradiation of tumors, I addressed the problem of inhomogeneous dose within a tumor. And, in order to have the model conform to experimental data in the limit of uniform irradiation, I introduced variability of cell radiosensitivity between supposedly similar tumors. The model had two surprising consequences. It predicted that the TCP would be only slightly, and not catastrophically, reduced by a relatively small dose reduction to a relatively small volume of the tumor. And, in a complementary manner, it predicted that a boost dose to most, but not all of a tumor could increase the TCP significantly; it would not be wasted dose leading to increased normal tissue damage but not improving the TCP. These predictions were quite different from what was then conventional wisdom, and I was very concerned that I might be missing something important. I was very relieved, then, after I spent a nervous hour presenting my model to Mort Elkind when he happened to be visiting our department, to find that he didn’t see any fatal flaw in my reasoning. Indeed, I had the feeling that he bought into my ideas.
I had a somewhat amusing and instructive experience regarding the publication of this model. I submitted a paper on the subject to the Red Journal. It was rejected on the recommendation of one reviewer, who wrote that my work was hypothetical and too speculative. I put the rejected paper in a desk drawer for several years, during which time I presented my model at numerous conferences and included it in the report of the first NCI working group which I have already described. I then decided to rework the paper a little and resubmit it. This time it was rejected, I am pretty sure by the same reviewer, on the grounds that it described things that were well known already! Publishing is, in truth, a very chancy business and one has to be a bit philosophical about it. As I’ve already mentioned, I tried in my career to keep my eye on the ball which is the cure of cancer. If one’s work advances that goal, then the issue of whether or not it is published over one’s name, or even whether one’s role is acknowledged, is largely irrelevant.
I think my model was an important underpinning to the treatments we were giving with protons at the Harvard Cyclotron Laboratory. Herman had the insight and courage to deliver treatments in which the dose to base-of-skull sarcomas was reduced in the part of the tumor adjacent to the brainstem in order to keep the dose to the brainstem below what was taken to be its tolerance level. The results of our strategy have been very satisfactory – and have given me, at least, some confidence in my model.
After I had developed it, I spoke at several conferences about modelling in general and my model in particular. But, how I addressed biophysical models depended on the audience. When I spoke primarily to physicians, I judged that they were generally (and quite appropriately) cautious about accepting any new approach. I responded to this by urging them to be open-minded. “There may be something to it.” I would urge. When talking mainly to physicists, whom I generally found to be overly enthusiastic about involving themselves in biological manipulations (remember my flirtation with TDF), I urged great caution and skepticism. These days, I have to say, I think caution and skepticism are needed across the board. Therapists accept a much greater degree of dose inhomogeneity in tumors than I think is wise, absent experimental proof that what they are doing is safe. And they accept hot spots in normal tissues with insufficient concern or, again, without experimental backup. I think people interpret the literature, to the extent they are aware of it at all, too superficially.
Dr. Suit: Well, so much for tumor control probability. What about the problem of normal tissue damage which you just hinted at?
Dr. Goitein: At the start of the first NCI-sponsored working group, which I’ve already described, I had successfully persuaded the group that we needed to be able to assess dose distributions in terms of their biological impact. And, in particular, to be able to compare two dose distributions in terms of the difference in their biological impacts and not the differences in the dose distributions. The issue then arose as to who would develop the necessary models. I was happy when John Lyman volunteered to take on normal tissue complication probability (NTCP) – in doing so, he made an important contribution to radiotherapy. This was around the time that Sten Graffman had tickled me to look into TCP, so I willingly undertook to take on the problem of estimating what happened when a tumor was inhomogeneously irradiated. I was glad to do that side of the problem because I had the philosophy that, when there was a choice, one should always tackle the simpler problem first.
I thought TCP was simpler for two reasons. First, the tendency with tumors was always to deliver an as-near-as-possible uniform dose so that, in practice, the non-uniformities were mostly modest perturbations of the dose – whereas, for normal tissues, recognizing that it is advantageous to spare as much as possible of an organ, the dose distributions were mostly deliberately highly non-uniform and hence it would be harder to estimate their biological impact. Then, the cells which made up tumors responded, I thought, fairly independently to irradiation – whereas the responses of cells within normal tissues and organs – one might now say functional-sub-units – could be highly correlated and we knew little about the quantitative extent and impact of such correlations. This second concern has, over time, heightened in importance in my estimation.
As a result of all this, I did much less work in the area of normal tissue complication probability. What little I did was as junior partner in a collaboration. And, here I would like to digress for a moment. I had the good fortune to have two gentlemen fall into my lap so-to-speak in the late 1990s. Andrzej Niemierko and Harald Paganetti. Both had obtained fellowships in their native countries (Poland and Germany respectively) and contacted me, asking if they could use them to come and work with me. In general I have a paradoxical attitude to free gifts; they are usually costly and I tend to refuse them. (I turned down an offer to receive for free one of the first Pixar graphics engines and, despite their subsequent huge success in the entertainment industry, never regretted the decision.) But both Andrzej’s and Harald’s accomplishments, to date, were persuasive, and to MGH they came. Since then I have had most pleasant collaborations with both of them.
I bring this up here because what little I did in the area of NTCP estimation was with Andrzej. And what might be of some general interest is the nature of the longstanding and sometimes energetic discussions we had about modelling. Our common ground is that the reaction of biological tissue to radiation certainly involves many complex processes. My preferred approach was, nevertheless, mechanistic. I thought one should try to identify the dominant processes and model them explicitly in the hope that the ones that one ignored might not be too greatly perturbing. Andrzej’s view, if I can venture to characterize it, was that we did not know enough to isolate the dominant processes and, in any case, had inadequate data to characterize them separately. In complex systems, he thought, one is best-advised to fit such data as one has with an empirical formula. To this day I think we continue to hold opposite opinions on this matter. However, I have to say that Andrzej has had more success with his approach than I have had with mine.
I’d like to add one thing. As I already emphasized, science (like haut culture) is subject to fashionable trends and, these days, classical radiobiology is out of fashion. Funding is hard to come by for it, and many of the old-timers have retired or given up. The emphasis is now on the sub-cellular scale. But, I am of the opinion that what I term ‘whole-organ radiobiology’ – the understanding of the response of organs and biological systems as a whole to radiation insult badly needs investigation. This is very important for guiding radiotherapy in an era where precise control over dose distributions and, in particular, the ability to sculpt partial organ irradiation, is now possible. Only, we don’t know what particular dose sculpture is best. We need data. Funding agencies please note!Dr. Suit: Let’s turn to proton therapy. You have been very active in that field. How did you get into it and what did you do first?
Dr. Goitein: Well, Herman, of course you were the guilty party. I started out in conventional X-ray therapy and fully intended to remain there, assisting in routine patient treatments and I envisioned doing research in that area. I had worked in particle physics for nearly a decade and I wanted to leave particles behind me. But then you were offered the opportunity to make use of the Harvard Cyclotron for radiotherapy and asked me if I would work with you, first to just look into the advisability of involving ourselves with proton therapy and eventually perhaps to develop a program. A bit reluctantly, because I thought that I had already found my niche, I got sucked in. You were never anything but persuasive and, of course, the superior depth-dose characteristics of protons made them very tempting. So, at first tentatively, I got on board. We went through several initial phases: a seminar with invited speakers; many internal discussions; the conduct of numerous radiobiological experiments both by you and by invited outside experts like Eric Hall, Raju, Paul Todd and Urano; and the building up by Andy Koehler of a treatment facility at the Harvard Cyclotron Laboratory. All this went pretty quickly because I came to the MGH in 1972 and we treated our first patient in 1974.
Our work on protons was a set of close collaborations between:
Dr. Suit: So, tell us about the role of each of these groups.
- the Harvard Cyclotron Laboratory (HCL), led by Andy Koehler;
- MGH medical staff – led by you, Herman;
- MGH physicists – led by me; and
- the Massachusetts Eye and Ear Institute (MEEI) – initially led by Ian Constable and taken over in the clinical phase by Evangelos Gragouds.
Dr. Goitein: The Harvard Cyclotron was a venerable machine, built just after the end of World War II. It already had a history of use for medical purposes, but our involvement was in the context of fractionated large-field cancer therapy and this was a new initiative for the laboratory. Andy Koehler was acting director of the HCL – his modesty did not allow him to call himself ‘Director’ although that is what he, in effect, was. Under Andy’s leadership, the HCL staff took on the job of developing the beam-forming and beam-delivery apparatus and, for decades thereafter, for delivering the treatments accurately. In that connection, Bernie Gottschalk played an important role as Andy’s scientific partner.
Andy had a particular style which permeated all of his work and strongly influenced his colleagues. He was at once modest, very knowledgeable, highly ingenious and meticulous in doing and documenting what was done. Most of all, his work was what I can only describe with admiration as ‘low end.’ What I mean by that is that his solutions to problems used the minimum needed to accomplish the job. Pencil and paper were far more common tools for him than computer print-outs. His many ingenious figures illustrated technical points perfectly. Small was indeed beautiful in Andy’s hands. His team emulated him in developing a warm informal style which permeated all that they did, their technical work, their seminars, their parties and, particularly, their caring support of patients. There were, in fact, two cultures: the HCL’s and MGH’s more structured and formal style. Mainly as a result of the disparities between these, while we worked effectively together, it was nevertheless not without its frictions, something that I always regretted but found myself unable to reverse.
You, Herman, of course, both headed the proton project and led its medical team. I say “of course” but it is by no means to be taken for granted that the chief of a large busy department would play so intimate a role in such a particular program. I think its success is directly attributable to your engagement in its development. You had some excellent helpers. John Munzenrider battled in the trenches for decades and Norbert Liebsch became a world expert in the treatment of base-of-skull sarcomas. Then we had the engagement of a sequence of more junior physicians such as Joel Tepper who was in at the beginning, Oscar Mendiondo, and later Jurgen Debus among many others.
After the preparatory work we treated three patients: a young boy with a rhabdomyosarcoma; a mother of two young children with a chondrosarcoma involving the petrous ridge; and a man with a mixoid liposarcoma of his thigh. This experience encouraged us to continue and expand our work which we did after a short hiatus needed to apply for and receive money from the NCI. We were very fortunate that they strongly supported us over numerous grant cycles and, indeed, continue to do so to this day.
I am fond of pointing out that some of our most substantial successes have been in the treatment of rare tumors such as sarcomas of the base of skull and ocular melanomas. No committee of wise men and women seeking to define a clinical program would come up with these tumors, both because of their rarity and because they are simply nowhere near the top of most practitioner’s mental lists. But, in fact, there are some real advantages to specializing in rare tumors. It is easier to establish a recognized center of competence in them, and one’s colleagues are not reluctant to refer the occasional patient to one since it barely puts a dent in their patient throughput (and, hence, pockets).
Although protons can take a great deal of credit for our successes, I did learn one thing over the years. Namely, that protons – or any other modality for that matter – are only a tool. The hand of the craftsman who guides the tool is a lot more important. So, when relatives and friends call me up about someone who has a cancer and asks me if they need protons, I always respond by saying that “No, they first need a really good physician with a lot of experience in that particular disease. Then, he or she will tell them what therapy or therapies would be most appropriate.”Dr. Suit: Let’s move on to the physics of proton therapy and your personal contributions.
Dr. Goitein: Well, of course, it was a team effort and Marcia and Lynn here were mainstays of that effort. Proton therapy would not be the same today had it not been for their involvement. But, I guess you’ll have to interview them to get their story. Parenthetically, it is a source of considerable discomfort to me that, because of their sheer numbers, I cannot name and give credit to all the people with whom I worked over the years. I owe so much to them.
As you obviously know, the design of a treatment system and then the planning and delivery of treatments involve a huge number of detailed tasks and activities. It’s impossible to describe the full list in any reasonably compact way but I’ll try to mention some of the larger items. One can identify individual physicists who took the lead in some of these things, but the fact is that we all had our hands deep into all of them.
Dr. Suit: Tell us about the treatment of ocular tumors. What was your role in it?
- Immobilization and localization Because the advantage of protons is their superior dose localization, and because the alignment of a compensating bolus with the patient’s anatomy can be critical, the localization of the tumor and nearby structures – and hence the patient – have a special importance and we put a great deal of effort into the design of patient support systems, techniques of patient immobilization, and methods to localize tumors.
- Dosimetry The dosimetry of the proton beam turned out to be difficult to tie down. Initially we depended on Faraday-cup dosimetry using a cup previously designed by Andy. However we, led by you, Lynn, also used ionization chambers and calorimetry and made a number of inter-comparisons with other centers. It took many years of effort to resolve the discrepancies which we found. Eventually we had to make an embarrassing 6.5 percent adjustment, I think it was, to our dosimetry.
- Horizontal CT unit Our work depended heavily on the availability of CT imaging. Proton therapy requires a map of the patient’s tissues with quantitative information about their density. CT evolved just in time to provide this map for us. It was important to us to be able to image the patient in the treatment position. Since we had no gantry, patients had to be treated either lying down or, particularly in the case of patients with head and neck tumors, seated. We needed therefore to be able to image a seared patient’s head and/or neck and this required a scanner that could scan a slice in a horizontal plane. The only glitch was, there were no such scanners. So, we had to persuade one of the companies to make one for us and, incidentally, the state regulatory agency to allow us to buy one. It was a long and difficult saga and I won’t go into it except to say that I formed the impression, à la Groucho Marx, that any firm that would consider building such a one-off device was a firm one wouldn’t want to do business with. Eventually, however, EMI agreed to build such a scanner for us. At the time they were the leading firm in CT, although they got out of the business two weeks after signing a contract with us! They built two other horizontal CT scanners, one went to the particle program at the Lawrence Berkeley Laboratory and I don’t know where the other ended up. Over the years this machine needed quite some care and feeding, but it nevertheless served us well.
- Impact of inhomogeneities When protons pass through matter they lose energy and eventually come to rest. They are also deflected slightly from their otherwise straight path though scattering processes. When the material traversed is inhomogeneous – that is to say, is composed of a complex of multiple materials such as bone, muscle, fat and air cavities, changes occur in both their penetration and in the paths they traverse. The two can and do interact in the sense that a change in path can involve a change in the materials encountered. When we started out these effects were, of course, known in principle but their quantitative impact on dose distributions had not been investigated. So we engaged in quite a long and demanding set of experiments and developed models which allowed one to take these effects into account. Perhaps the most startling perturbation that could occur was what we later termed ‘distal edge degradation’. This meant that the normally sharp fall-off of dose at the end of range seen in homogeneous materials, and on which the benefit of protons partially depended, could be substantially modified by inhomogeneities. The dose would then fall more gradually at the end of range, spreading sub-maximal dose over even several centimeters rather than the usual millimeters. We were apprised of this problem in an interesting way. Our fourth potential patient was a man with a nasopharyngeal tumor, referred to the proton group by CC Wang, the senior staff member responsible for managing patients with head-and-neck disease. The idea was to irradiate the patient from one side and bring the beam up to, but just short of, the contralateral neck nodes and skin. On the way, the beam would have to go through some substantially inhomogeneous regions. I was concerned as to whether we could pull this off, and we designed a test in which we would irradiate the patient using a beam whose energy was increased over that which would be needed for the actual treatment, thereby adding some known centimeters of range to it. We built a ‘telescope’ consisting of a series of several thin plastic sheets interspersed with films. By exposing this telescope to a tolerably low dose of protons we could, after developing the films, determine where the protons stopped in any particular part of the field. By calculation we knew where they should stop. Had the patient been composed of uniform matter, the protons would stop sharp so that one film would be darkened and the following film not. To our surprise and consternation there were areas of the field where the darkening took place gradually, over several layers of the telescope. We had seen distal edge degradation. CC, not without some energetic ribbing of any of those of us he encountered over the next days, took the patient back and treated him with X-rays, and we launched our program of investigation into the perturbative effects of inhomogeneities. You, Marcia, demonstrated this effect in some very elegant experiments. So impressive was this experience that, to this day, I remember that patient’s first and last names as, indeed, do I recall the names of those fist three patients we treated using protons.
- Treatment planning I have already talked in extenso about our development of 3-D treatment planning. I just want to add that writing the programs was just one side of the problem. Enormous effort went on over the years to plan the treatment of each patient. Lynn and Marcia know all about that; they were in the trenches. I want also to mention Judy Adams whom I met at a conference in Denver and immediately recruited to MGH where she has been a pivotal person in the treatment planning effort ever since. Along the way we evolved several strategies for solving difficult geometric problems. Field-patching was one of these, in which we used two or three beams, coming from different directions, to cover the tumor volume, none of which individually treated the entire tumor but which together did so, and we ‘feathered’ the beams where they abutted to reduce the magnitude of hot or cold dose spots due to possible small misalignments. This was, in effect, a crude form of intensity modulation, developed long before intensity modulation got its name.
- Study design and implementation For completeness, and because it often involved a lot of effort, I want to add that we physicists participated, of course, in designing and then implementing the various clinical studies which were undertaken.
Dr. Goitein: Initially, Milford Schultz, who preceded you as radiation therapy department chief at the MGH and, I believe, Dr. Regan of the Massachusetts Eye and Ear Institute had the idea of using protons to treat childhood retinoblastomas. They involved a young Australian ophthalmologist, Ian Constable, in their scheme and he quickly redirected the effort towards the treatment of ocular melanomas for which he thought protons would be particularly well suited. His involvement took place at the time that you, Herman, had already taken over as department chief and you roped me in to help Ian perform experiments on the eyes of owl monkeys with both protons and 2 MeV X-rays. This led to a clinical program for the treatment of, mainly, choroidal melanomas. As usual, Andy Koehler designed and built the beam delivery system with his usual ingenuity. He held the philosophy that it was better to build purpose-build devices for particular problems, rather than general devices to handle all-comers. In large part I have come round to his opinion.
I designed the software program for planning the eye treatments. This was developed before the large-field program that I have already described and it used quite primitive hardware. Because the eye has such a homogenous density, one did not need CT and the imaging of the eye and the embedded tumor was primarily based on ultrasound measurements. The program turned out to work well. The fact that local control rates of 95 percent and better are obtained in treatments based on that program attest to its satisfactory function as well as on the adequacy of the treatment technique that we developed. The treatment of choroidal melanomas with protons has spread widely and there are some 14 centers at which proton treatments are or have been given at present counting. The MGH/MEEI collaboration treated for many years the largest number of such patients but by now the Paul Scherrer Institute in Switzerland, whose medical program is presently headed by my wife, Gudrun Goitein, is the world leader. My program has proved to be quite robust and the majority of the several centers now treating choroidal melanomas with protons still use it – although, over the years, some very useful enhancements have been added by others.
Dr. Urie: Michael, I think one of the most significant contributions that you made was in the area of treatment planning, not only in proton therapy but also in the X-ray world. You hinted that one of the regrets you have is that your confidence level approach has not been incorporated widely. Where else do you see treatment planning needing improvement?
Dr. Goitein: Well, I've been a bit out of the scene for probably the last 15 years, so I'm not a good commentator on the current situation. What I would say is that there is a big disconnect between what is commercially available and what the best people in the field are thinking and doing. Now, maybe there is always something like a 20-year gap for innovations to take root and we get there eventually. But the development of treatment planning programs has become really quite expensive for the commercial companies and they, by-and-large I think, do as little development as they can get away with. Institutions which are new to proton therapy and are in the market for treatment planning programs tend to be relatively unknowledgeable about the finer details of what is needed, so they are not in a position to use the power of their purse to nudge the companies into necessary new developments such as, for example, including confidence limits. The community might do well to form an expert group to push the companies to do more.
I can think of three areas (over and above confidence limits) in which improvements would be desirable. First, streamlining the user interface. A user should only have to do things that only a human can do; the rest should be automated. One ‘clicks’ and types far more often than is necessary and that is simply a waste of the user’s time and competence. Programmers, given the chance, will write the code which is easiest for them to write. They have to be persuaded to be the user’s friend.
In the field of proton therapy the issue of robustness is now receiving attention in academic centers – Thomas Bortfeld’s group at the MGH is a leading player in this effort. Robust planning is the art of avoiding geometries in which small errors can lead to large problems. The development of techniques to assure robust plans has not reached the commercial world to any great extent but it is very much needed.
My third area in which developments are sorely needed is really in the field of radiobiology – although its solution would then have to show up in treatment planning. What we have learned in the last 20 years is that we have the tools now to tailor an inhomogeneous dose to both the tumor and normal tissue. I think the tumor side of the problem is fairly OK. But on the normal tissue side, we really don't know how to shape the dose within, say, a kidney or brain or whatever in order to produce an optimal plan, all things considered. Physicists and computer people could implement solutions very easily if we knew what they were. But we don't. So I think one of the big problems is that we don't have any very substantial investigation going on into learning how best to tailor dose so as to minimize organ damage. Recently there was a series of reviews of what we know under the acronym QUANTEC, and I think that it has been a valuable contribution, but it has also highlighted how little we know. A particular area of ignorance, which is important in the context of proton therapy, is in knowing what the implications are of the additional dose bath which X-rays deliver as compared with protons. We have no idea of what the clinical implications are of reducing the dose from 35 gray to 15 gray over a large volumes. We desperately need really long-term studies to try to look into dose-volume effects. So I think there is a whole area of radiobiology that has come up as being important and valuable which is not getting attention and is not getting funding. And, as I already mentioned, the lack of funding of traditional radiobiological research has meant that many of the experts have left the field.
Dr. Suit: All the activities you describe needed quite a bit of money and you were responsible for putting together most of our grant applications. Was that a pleasant experience?
Well, I did accumulate quite some experience with grants. Over the course of my career I was responsible for putting together something like a dozen grant applications for NCI funding. I think that, for a short while, due to the happenstance that I was the principal investigator for the grant to build the MGH’s proton center, I was the top MGH awardee of NIH grant money.
Such grant applications are typically at least a couple of hundred pages long. They have to be carefully written and well produced. Some people resent the effort that this entails, but I never did. Writing a grant application forces one to assess what one has accomplished, to think carefully about what one wants to do in the future, and to justify it convincingly, first to oneself and then to the reviewers. This effort is good for the entire team and, moreover, it is reasonable that one be required to do it. I learned a few tricks of the trade, foremost among which is that one should read the instructions carefully and then follow them – no matter how unnecessary or foolish one thinks them to be.
I was always impressed with the seriousness and integrity of the grant reviewers and the grant administrators. I, myself, was a member of site-visit committees for several large grant applications (so-called program-project applications). These were very serious affairs which were intellectually challenging and from which one learned a great deal. I got, I think, a reputation as a tough interrogator. I always felt one should ask the difficult questions. I know for a fact that I was blamed by unsuccessful applicants of two large and important projects for their applications having been turned down. In fact, although I certainly posed difficult questions in the open sessions, in the end I supported the applications in both of these two cases and it was other committee members who nixed them.
Dr. Suit: You mentioned the grant to build a proton center at the MGH. How did that come about?
Dr. Goitein: Well, there was an external and an internal basis for wanting to build a new proton therapy facility at the MGH. Externally, the dean of Harvard’s faculty of arts and sciences woke up to the fact that the faculty was hosting a medical program – namely the work at the HCL, while having no medical competence themselves. So he wrote a letter, I think to MGH’s director general, asking what his plans were for terminating the HCL program as Harvard had other uses for the land. This letter, of course, put a lot of things in motion. We could surely have resisted this implicit request to move, but we had our own reasons for wanting a new facility and so we’re happy to have this pressure applied. The HCL, while a truly amazing facility, had some fundamental limitations: the cyclotron’s energy (160 MeV) was too low for its protons to reach all but relatively shallow tumors; there was no room to build a gantry which was needed to allow treatments from any direction; and the site was distant from the main MGH radiation oncology department and from needed hospital services such as radiology and anesthesiology with the consequence that the proton program was a bit cut-off from the mainstream activities of the department and the hospital.
So, we saw Harvard’s initiative as an opportunity to improve the program. I will not go into the details of how we went about that, it was a long and complicated process. But I will mention a few high points. However, I first want to mention my first step. I knew that this would be a big job and I did not want to embark on it unless I was assured of the clinical appetite for it. I had seen too many physicist-driven clinical projects which languished because they did not have enthusiastic medical support and involvement. So I went to you, Herman, and I told you that I was willing to help you with the project, but that the decision to do or not to do it had to be yours alone. It had to be your project and I would not venture an opinion on its merits. Well, you bit and the rest is history.
All that was needed was a space in which to locate the new facility and the money to build it. Anyone who has worked in a large urban hospital knows that, of these two needs, finding space is often the more difficult to fulfill. However, we were fortunate in that regard. The MGH, in some sort of a land swap with the city, had just acquired the Charles Street Jail which abutted the MGH campus. We got to put the new center in the old exercise yard of the jail. That left us with the money problem. We applied to the NCI for permission to apply for construction funds and they turned down our request on a technicality. We had no option then but to lobby congress for funding. However, and I was very happy for this, the MGH had a policy of not applying for so-called earmarked money funds that were allocated to a specific institution. What they would do was to lobby for money to be allocated to build one or more proton therapy facilities in the US and we would then compete with all comers, with luck successfully, for a piece of the pie.
I do feel the urge to relate one anecdote. I had estimated that the facility would cost in the range of 40 to 56 million dollars. (This was based on a wild under-estimate I had received from the Loma Linda people who were just then building a proton center which, in the end, cost them in the neighborhood of 100 million dollars.) We had obtained a meeting with the president’s science advisor of the time, Alan Bromley, himself a physicist. A few of us trooped down to Washington to make our case. We gave him a figure of 120 million dollars for three facilities to be located nationwide. Several months later I met up again with Dr. Bromley at some event taking place in a fancy Boston hotel, and I reintroduced myself to him, saying that he might not remember me but that I’d been in his office, pitching for funds for proton therapy. “Remind me, how much were you looking for?” he said to me. “One hundred twenty million dollars,” I replied. “Michael,” said he, “you don’t really think that I micromanage the budget, do you?” Thus it was that I gained a personal perspective into the size of the US federal budget.
To cut a very long story short, our lobbying was successful. We went through a two year planning phase with our one competitor, a group from San Francisco. The NCI decided to fund just one facility and, in a face-off with our California friends, we got the nod. So, now we had both space and money. All we had to do was to build the facility.
Let me add one thing. I just mentioned the Loma Linda proton facility. This was the brainchild of Jim Slater – the first hospital-based general-purpose proton treatment center to be built, anywhere in the world. It was a ground-breaking achievement from which we at the MGH learned a very great deal. Our facility, both in its conception and in its design, owed much to Loma Linda’s pioneering work.
Dr. Suit: And, how did you go about building the facility?
Dr. Goitein: You know, one could write a whole book on that subject alone. I’m hard-pressed to know what and how much to say about the construction process itself. It lasted a long time; from 1993 to at least 2002, and some would say, I among them, that it’s not finished yet. And, it was arduous. I’ll try to list out just the main highlights.
- A firm cost for the project had to be established and the money secured. In the end we had 6.1 million dollars from the NCI for a planning phase and then 40 million dollars for construction, half from the NCI and half from MGH funds. I am proud of the fact that, with a lot of help from my friends, I brought the project in within budget – although not, alas, within schedule.
- I built up MGH’s core project staff. My goal was to have people so experienced that they could stand toe-to-toe with our contractors’ staff in any discussions. The team consisted of five besides myself: Al Smith as deputy director cum medical physicist: Stan Durlacher as construction expert; Jay Flanz as accelerator physicist; Anne Levine as hospital planning expert; and Susan Woods as grants manager and general problem solver.
- The next stage was to decide on the overall strategy. We decided to split the job into two pieces purchase of the proton therapy equipment and of the building and to let two separate contracts, one for each piece.
- We then wrote up specifications for the project. So far as the equipment was concerned, I decided on a novel approach. Instead of telling the provider what to build (e.g. a cyclotron of such-and-such an energy and intensity), we told them how the system as a whole should function (e.g. it must deliver a beam of a stated range of sizes and range of penetrations in the patient from any angle and with at least a stated dose rate). This made clear to the provider what he should accomplish – i.e. the nature of the radiation that a patient could receive – but left the technical design up to the vendor – e.g. the accelerator could be a cyclotron, a synchrotron or even some other device. The approach of writing functional specifications has since been adopted by almost all subsequent proton therapy facility customers.
- For both contracts, after identifying qualified interested bidders, we published a request for proposals and then, after evaluating the bidders’ written responses, we gave them an opportunity to present their proposals orally and to answer questions. In all this we relied heavily on a panel of outside experts for advice.
- We then identified the preferred provider for each contract and negotiated, and eventually signed, a contract. This was a long and expensive process, involving as it did some very pricey lawyers, which lasted most of a year. But, it was time and money well spent in my judgment. The execution of the work is greatly facilitated by a clear and detailed contract without ambiguities. Recently there has been more than one proton facility whose construction has got into great difficulties through not having a clear contractual understanding between the parties.
And then, the work began. In the end, Ion Beam Applications (IBA), a Belgian firm, was selected to design and build the proton therapy equipment and the Bechtel Corporation to design and build the building and to support MGH in some management matters. The biggest problem with the decision to have two separate contracts with two separate providers was the danger that they would simply point the finger at the other if any problems arose and, in the worst case, would be at loggerheads. We dealt with this in part by building into the schedule a coordination phase in which IBA had to develop an interface document which specified everything that they needed the building to provide. Developing this document was a long and difficult matter, but it paid huge dividends in the end.
Time passed and the project progressed. So far as the building was concerned, things went smoothly. The building was designed and built within the scheduled time. I learned that time is very much money in the building trade. I had built a contingency of something like 2 million dollars in our budget for the building phase. I needed to conserve it in order to be able to buy a second gantry for the facility – which was an option in our contract with IBA. With Bechtel’s willing support, we were able to protect the contingency fund and, as a result, we were, in the end, able to buy the second gantry.
The equipment side of things did not go so smoothly. We had divided their deliverables in several parts and payment for each part was contingent on IBA passing a review by the MGH aided by outside experts. Each review typically lasted two days. In the end, we held some 20 such review sessions. On the technical side there were a few substantial problems, wear of the gantry’s rolling surface, for example, which cost time but which were all manageable. More seriously, as they came to assemble the whole apparatus and test it, numerous small problems cropped up and it seemed as though for every problem that got fixed a new one cropped up.
The most devastating technical problem was with the development of the equipment control system. This problem was responsible for much of the eventual delay in completing the equipment contract. IBA had badly underestimated the difficulty of developing the control system. Once the problem was realized, IBA scrapped their in-house development to date and sub-contracted the entire job to an outside company. This cost them at least a two-year delay –and a whole lot of money which, as we had a fixed price contract, they had to swallow.
With all these problems, everyone was getting nervous: IBA, the MGH directorate, my team, and, not least of all, I myself. I reacted by dividing IBA’s deliverables into two parts, only requiring the first de-scoped part to be delivered initially. This was disappointing, but it was an essential step; we would never have got to patient treatments otherwise. As a result, three years delayed from the contracted finishing date, IBA was able to deliver a somewhat incomplete but nevertheless functioning system.
Dr. Suit: This was a large multi-million dollar project. Did you have any previous experience with such a large undertaking and, if not, how difficult was it for you to manage it?
Dr. Goitein: No, I really did not have any relevant experience of management on the scale that was involved. Like many academics, I had a rather disparaging view of ‘management’. My technique had been to lead by example and, when the going got tough, to implore my co-workers to “just settle down and do the job.” This does not always work, especially in a large and complex project, and over time I came to appreciate the importance of good management and formal procedures. In this I was tremendously helped by Paul Reardon, a dear man who is now sadly dead, who became a consultant to the project. He held my hand and cheerfully educated me in these matters.
I learned that good management requires that one has: clear and adequate specifications, a schedule and a budget. One then needs to keep within the schedule, to keep within the budget, and to achieve the specifications. One has to allocate both responsibility and authority to one’s co-workers and ensure that they have the resources needed to do their job. All this needs to be documented, and one must audit people’s work to make sure they are following the plan and that the work is of good quality. That is all. I wish I had known that from the beginning.
I learned one other useful thing from a little story attributed to the French author André Malraux. The bishop had reached the ripe age of 80, and his congregation put on a celebration. A young journalist was dispatched by the local newspaper to interview the great man. “Your Excellency,” he asked, “you have certainly met with many people in the course of your ministry. Can you sum up for my readers what your experiences have taught you?” The bishop paused a moment for thought and then spoke. “Well, young man, I have learned really only one thing. There are no grownups.” So many management problems, I realized, come down to the immaturity of the people involved.
Dr. Suit: Michael, you left the project before it was completed. How did that come about and was it difficult for you?
Dr. Goitein: Well, what was difficult was remaining with the project as it dragged on. My early departure came about for personal reasons. I had met, fallen in love with and, after three years, married Gudrun, a wonderful and exceptional woman if I may be allowed to say so. She lived and worked in Switzerland and, for both professional and family reasons, we decided that I would move to Switzerland rather than she to the US. However, I felt that my presence in Boston was essential for the success of the proton facility project, so I remained there for four years after our marriage, traveling to Switzerland for about a week each month. The more than seven year separation was, of course both physically, because of the travel, and personally demanding also for Gudrun. I waited until I judged that the project was finally coming together and that Flanz would be able to oversee its completion, and then I resigned from the MGH and moved to Switzerland. I think that the results demonstrate that my judgment in this matter was correct.
Dr. Suit: So, what did you do, once you had moved to Switzerland?
Dr. Goitein: Well, I had to decide whether to look for a position doing pretty much what I had been doing for the past thirty years, or whether to try something different. Once I put it to myself in those terms, the choice was clear. Something different. I decided to become a consultant, primarily in the field of proton therapy. That might not seem different at all, but in fact, as I learned, a consultant’s tasks and perspectives are quite different from a researcher’s. I decided to restrict my consulting to supporting people who wanted to set up a proton facility, rather than to consult for industries who wanted to sell them. One cannot do both, of course. That would lead to enormous conflicts of interest. I set up a small one-man company for liability protection reasons, and went to work. Let me briefly describe a few of my projects.
A group in the French-speaking part of Switzerland were considering raising a proton therapy project and asked me to consult for them. I mention them for the paradoxical reason that I helped them decide that they did not want to pursue this plan. I considered my job to be to assist my clients in first learning, and then achieving, their objectives. In this case, the people did not really have an appetite, nor did they have a sufficiently large patient base, for a proton project, and I helped them realize that. So, although my relationship with them was short, about one year, and no facility resulted from it, I nevertheless considered it to have been very successful.
My most substantial project, which lasted almost ten years, involved assisting a group in Trento, in Northern Italy, to raise a proton facility. The “group” initially consisted of one man, Renzo Leonardi, who conceived of the project and has carried it forward all the way. I was persuaded to work with him for two reasons. First, he, and later the group, really wanted my advice. I can’t tell you how rare that is. Mostly, people want to hear you confirm what they intended to do in any event. Second, the funding of the project was to be provided by the Province of Trentino and, in effect, there was no business plan. It was not intended to be a money-making project. It probably didn’t even have to break even. The idea was that the province was providing a benefit for its citizens. How refreshing! This, too, was a rare approach, and one that greatly appealed to me. So, I helped Renzo to: sell the idea to the Province; build up a project team; define how the facility should look through developing detailed specifications; go through a procurement process very similar to the one we went through at MGH; select a provider; and conclude a contract with IBA, as it happens. I then elected to leave the project. It had been a very fruitful collaboration, but I felt that the Trento group needed to stand on their own feet and not rely any more on my prodding and advice.
Finally, I will mention a quite different relationship. Radhe Mohan, at MD Anderson, asked me to consult with him to help him define a research program and then write an application for NCI funding. We had a very pleasant collaboration over several years. I persuaded him to join forces with the MGH in seeking funds and in conducting the research, which he did, I think, to their mutual advantage. At least they were successful in getting funding for their research.
I had thought that becoming a consultant would have the benefit of allowing me to taper down my level of effort over the years and, indeed, that was so. Eventually, in 2011, I decided to retire completely. I gave up my consulting work and, indeed, all professional activities, no more articles, no more talks, no more conferences and so forth. Or, at least, I tried to do so. It is amazing how many things keep cropping up which it would be churlish to refuse to do. But, I’m getting there. I find that the moment I have even the most minor professional obligation, it occupies my mind virtually full-time. I don’t want that anymore.
Dr. Verhey: There is something that Michael hinted at but didn't really talk about directly and that is the use of sledge hammers to drive small nails. The bottom-line aspect of particle therapy, in particular proton therapy, means that there are places treating prostates with two proton fields because there are a lot of patients who want it and can fill their coffers. But, at least as I learned with you guys, that's not the best use of protons. So what are your thoughts about, let's say, the use of protons for something common like prostate cancer which, while they won't be worse than using X-rays, might in fact not be significantly better. Shouldn’t protons be reserved for situations where their advantage is much more obvious?
Dr. Goitein: Well, I think that that is certainly the toughest question of the afternoon. It raises a number of difficult points and I’ll try to address them though, since time is short, only briefly. Perhaps I should start with cost since most informed persons agree that, were proton therapy cheaper than, or even as cheap as, X-ray therapy, protons would be preferred. Protons are, without doubt, more costly than X-rays. Having said that, I have to remark that the charges for proton therapy in the United States are, in my opinion, way out of line. They are, for example, between 5 and 10 times higher than the comparable charges for heavy charged particle therapy in Europe. This means that proton therapy revenue in the US must greatly exceed its cost and this, as your question implies, is a strong incentive for a proton facility to use its protons on as many patients as possible. This incentive has two consequences: patients whose benefit from protons is only modest, but whose treatments are simple and hence rapid, may be preferred over more difficult cases; and there may be an incentive to take short cuts in designing the treatments so as to make them simpler and faster (e.g. treating only one field per day). Prostate treatments share both these aspects.
With regards to these points I want to make a few observations. First, one has learned some radiotherapeutic principles from conventional radiation therapy, and they haven't changed just because one is using protons. For example, we have learned that treating all fields a day is kinder to the patient’s normal tissues than is one field a day. One-field-a-day proton therapy is being used purely for financial gain. The interest in hypo-fractionation is also relevant in this regard. I wish that the appetite for hypo-fractionation came from a belief that it was advantageous to the patient rather than from the fact that it makes a treatment course cheaper. Let's do clinical trials and find out – here I do think that randomized trials could be ethically designed. But to jump in and treat with fewer larger fractions because you can get more patients through and can get away with it is, I think, very undesirable. Very. To me, it is madness to build an expensive facility and then degrade its performance to reduce costs. Lastly, as most of my colleagues are well-aware, the costs of radiation therapy, even with protons, are very modest as compared with those of many other treatments. I cannot sympathize with the fact that the critics of proton therapy on financial grounds do not first turn their guns on, say, some very expensive regimens of chemotherapy.
However, given the higher cost of proton facilities and their relative scarcity, it makes sense to limit their use to clinical situations in which conventional X-ray therapy is unsatisfactory – either because local tumor control is inadequate or because complications are either too frequent or too grave. Prostate treatments would likely not meet these criteria. However, I must sound a note of caution. The advantage of proton therapy may often lie in the reduction of late effects. These complications may become evident years and even decades after treatment, so very long follow-up times are needed to evaluate protons’ performance.Dr. Suit: Before we wrap up, as you bring the subject up, I want to ask you your thoughts about the ethics of randomized clinical trials. I know that you were engaged in this matter in the last few years. Can you tell us a bit about that?
Dr. Goitein: For many years I was intimately involved in the design and the practical conduct of our research protocols regarding proton beam therapy. This necessarily put me in the middle of the controversy, which rages on to this day, as to whether randomized clinical trials (RCT’s) are appropriate for evaluating proton therapy vis-à-vis conventional X-ray therapy – and my opinions on the subject have matured over the years. A couple of years ago, I collaborated with Jim Cox to write a paper on the subject. It was published in the JCO and, I believe, attracted a record number of editorials and letters to which we offered a lengthy reply. Subsequently, Joel Tepper, Anthony Zietman and I wrote a somewhat related piece on what we termed ‘technology evolution.’ So, let me just mention the main established and controversial points with respect to proton therapy.
First, some experimentally supported facts:
- for a given tumor dose, protons almost always deliver less dose to uninvolved normal tissues;
- normal organs and tissues are damaged by radiation and, in virtually all cases, the damage is greater, the greater the dose; and
- protons do have additional problems relative to conventional radiations (especially as regards the impact of tissue inhomogeneities on their dose distributions) but techniques have been developed to satisfactorily overcome these problems.
These considerations make it quite unlikely that proton therapy would be worse than conventional therapy (provided comparable levels of technology for dose-delivery are used).
To justify a RCT, ‘equipoise’ must obtain – that is, the arms of the study must be judged to be comparably good. This is not controversial. But, the devil is in the details. Three issues are not agreed upon:
- what evidence is appropriate in judging the relative goodness of the competing therapies;
- who should make that judgment; and
- at what confidence level should equipoise be judged not to obtain.
Briefly, RCT advocates think that only clinical data are relevant (in a cunning bit of double-speak which George Orwell would have appreciated they have coined the term ‘clinical equipoise’ and substitute it for ‘equipoise’ in their definition of the ethics of RCTs) whereas I think that all existing relevant data should be considered (including, in the case of protons, the physical and clinical data mentioned above). Moreover, RCT advocates think that investigators’ judgments, backed by that of an Institutional Review Board, is all that is needed to justify an RCT, whereas I think that a patient’s likely opinion (informed by all existing relevant data) should have a central role.
There are a couple of terms whose use in connection with the ethics of RCTs can be misleading. The first is embedded in the statement that “We do not know that protons are better for the patient than X-rays.” This is sometimes stated by proponents of RCTs as a justification for judging that equipoise obtains. But the issue is not what we know with certainty, it is what the probability is that protons are better. And, while academicians would like to use a very high standard for probability – 20:1 for example – patients are likely to be content with a much lower threshold – 2 or 3:1 for example. The other misleading term is embedded in the statement “There are uncertainties surrounding proton beam delivery” which is taken to imply that, since there are uncertainties, equipoise must obtain. Well, of course there are uncertainties – but what matters is their magnitude. Are they great enough to likely make the use of protons inferior? This is a technical judgment and generally those knowledgeable about the technical aspects of proton beam therapy judge that the uncertainties can be managed (e.g. with safety margins) to the point that the level of any uncertainties is acceptably low.
So there, in a nutshell, is the argument. I can sympathize with the desire to obtain the best possible data, but one has to accept that some things which are scientifically desirable can simply not be done on ethical grounds. Most practitioners of proton therapy hold opinions similar to those I have described as being my own and, as a result, there have been almost no RCT’s of proton beam therapy. The hidden agenda is cost. If protons were cheaper than X-rays there would be no discussion. Unfortunately, they are not.Dr. Urie: I'd like to ask Michael if he would expand his views on the role of statistics and statisticians in the radiation oncology world?
Dr. Goitein: Well, so far as statistics are concerned, I have emphasized the importance of uncertainty analysis in almost all areas of radiation therapy (and, indeed, of life) – and it goes without saying that statistics provide the tools for such analyses. In fact, I have long felt that we should teach statistics to school children from a very early age.
Statisticians are another story. My opinion is that they are necessary in radiation therapy but not sufficient. They are necessary for two reasons. First, clinical investigators are often surprisingly naive about the demands of statistics on the design of clinical trials; they can wildly underestimate the patient numbers needed to obtain the desired power and overestimate the gain that they hope to see. A statistician can help them here – and should be involved from the very start, not at the last moment to ‘fix up’ a flawed grant application (as they legitimately complain is frequently the case). Then, too, even if the investigator is statistically competent, it always helps to have a second pair of eyes look over one’s experimental design and over its implementation. So, I think that, at the Department level, the involvement of a trained statistician in any proposed clinical trial should be made obligatory.
However, in my mind, statisticians are not sufficient, also for two reasons. First, professional statisticians tend to be focused on a rather narrow range of problems – the design of RCTs in particular. They have the tools for such problems and are good at using them. But I have found that they tend to waver when a problem does not fit into their armamentarium. And, radiation therapy is full of ‘unconventional’ problems – for example, extracting a model for the dose-volume effect in an organ from ‘dirty’ data, or measuring (with uncertainty bounds of course) the magnitude of an effect rather than testing a hypothesis. Second, with regard to RCTs, they are no more competent to make judgments about the ethics of a proposed trial than you or I. And the worst of it is that statisticians have the bias that RCTs are the gold standard and hence have the urge to fit all questions into that mold. We, the clinical team, must take responsibility for the ethics of our proposed trials and not be intimidated by those who may have ethical blinkers.Dr. Verhey: As we get to the end of our time, I would love to hear you opine on what you think were the contributions that you have made to the field of which you are most proud. I think they're all in here, but maybe if you could sort of point a finger at some of the things that you think have made the biggest difference in the treatment of patients with protons or X-rays, for that matter. I would love to see that in print.
Dr. Goitein: Well, that is an embarrassing question - but I confess I do have a few thoughts on the matter. I’d like first to say that almost all of what I did in medical physics was not complicated physics and certainly did not involve complicated mathematics. It was purely applied common sense. I think that, had I not gone into medical physics, others would have done most of the same things in due course. Perhaps I bought a few years on some of these developments. Interestingly, many of the things that I developed, almost always with the help of colleagues – yourselves, Marcia and Lynn, in particular, and often based on concepts mooted by others – have seeped into general clinical practice and no one knows or cares anymore that we were behind them. In fact, I find that very comforting.
If you twist my arm to make a short list, I would include:
Dr. Verhey: Unfortunately I have my own answer to my own question because I think the thing that makes you so unusual and important in the field is that you always ask the question of what's the most important thing in terms of solving a particular patient's problem. What are the things that are going to cause you to potentially make errors in what you think you're doing? Which area of the treatment do you have to worry about the most? Is it the patient immobilization? Is it the definition of the beam? Is it the fractionation schedule? Whatever it is, you always were asking the right questions. So, unfortunately my answer to my own question is that I think you were the only person - I'm not sure I ever ran across anybody else, I hope I imitated you - but I'm not sure I ran across anybody else in the field who ever thought as logically or with as much common sense as to how to look at every patient's problems with a physicist's point of view and find the best solutions.
- analysis of the effect of uncertainties on planning and delivering radiation therapy. This was something that I pushed quite hard and I have to say it's been an uphill battle. It's been very disappointing to me that uncertainty analysis has not become more ingrained in the field. However, some of the work that is going on now under the rubric of robust treatment planning is to a large extent directed towards reducing uncertainties, so that's a very good development in my estimation.
- development of 3-D treatment planning for proton treatments of choroidal melanomas;
- measurement and analysis of the effect of tissue inhomogeneities on proton dose distributions;
- development of image-based 3-D treatment planning for both X-ray and proton therapy – what is now termed IGRT. Novel developments included: incorporation of CT imaging, beam’s-eye-view, digitally reconstructed radiographs, dose-volume histograms developed with you, Lynn, display of uncertainty bounds on dose distributions, side-by-side display of color-wash dose distributions for rival plans and so forth;
- facilitation of non-coplanar treatments. Since the time of introduction of isocentric therapy machines, virtually all treatments were delivered using coplanar beams. Our program and the use we made of it completely changed perceptions or what was possible so that non-coplanar treatments are today freely employed;
- introduction of simple aids to improve patient alignment, for example; thin foam mattresses and perforated thermoplastic masks and body molds. Today you can hardly go into a radiotherapy department and not see these masks lying around all over the place. That's something I feel very good about although almost no one knows that I originated these things;
- development of biophysical models of TCP and, to a lesser extent, NTCP – and their use in guiding treatment planning and informing judgments about rival plans;
- guiding the building of a commercially-developed hospital-based proton therapy center at MGH based on a set of functional patient-driven specifications.
- In later years, contributing a bit to the discussion of the ethics of randomized clinical trials for particle therapy.
Dr. Goitein: Those are very warm words for me and I thank you for them. Common sense is a surprisingly rare commodity. You, Marcia and Herman all have enormous common sense and that was very important for our work together.Dr. Verhey: I think we learned it from you, Michael. I don't know what Marcia would say.
Dr. Urie: I think it comes in my genes, too.
Dr. Goitein: I suppose the other thing is a patient-oriented view of things. Very often somebody who will come in and ask “What can I bring to the table?” Whereas, when you're tending to patients, the right question is “What does this person need?” And I think we had that orientation. Our weekly proton rounds were very focused and very exciting because we would ask just that sort of question. When I'd visit another hospital, I would ask to attend rounds, and most of the time they could be pretty boring. You know, images and plans were thrown up and everybody sort of nodded because they were half asleep and then they would go on to the next patient. I think our rounds were very energetic and interesting. The rounds at Harvard’s Joint Center were the same way. I went over there a few times and Sam Hellman really stimulated a lively scene there. I think this level of active engagement on the behalf of the patient was quite an important element of what we did and was very - it was a big part of our success.Dr. Verhey: Yeah, and I don't know whether it's attributable to the institution or to Herman or to you, but I think the lesson I learned there is that the way you make treatments better is you bury yourself in the clinic, identify patient-centered problems and find solutions that can benefit future patients as well as the particular one under discussion. I know lots of leaders who wouldn't basically remember where their clinic is. You were there. We all were there. That's where I think the advances came from: solving clinical problems on a patient-by-patient basis.
Dr. Urie: I would agree with that.
Dr. Suit: Michael, is there anything you would like to add?
Dr. Goitein: I would like to add a couple of remarks. I bemoan the fact that the world of radiation oncology and, indeed, of medicine in general has, over the period that I have been involved in it, become far more driven by the bottom line, by considerations of profit margins. And, what I particularly regret is my impression that the soldiers in the field, rather than resisting the administration’s pressure with all their might, have not only accepted it, but have fallen in line to support the new model. For example, in my opinion, in proton therapy compromises in treatment technique and selection of patients are more and more made for financial reasons – and physicians are less and less inclined to represent the patients’ best interests.
And second, although I do not wish for immortality, I would like to come back for one day every, say, 50 years as a fly on the wall. I would like to find out how cancer treatment – and, indeed, how the world in general – has evolved in the intervening years. I’m not very optimistic. I don’t have a clue how it will change - but change it will, dramatically I would predict. I wonder whether the things with which we occupied ourselves during my professional lifetime will turn out to have ‘weathered’ well and how many of them will have become at best historical curiosities.
Finally, let me say that it was a great pleasure to hear your voices. It's been a long time. You know, I've recommended to virtually everyone who has told me that they are considering an offer to become chief of a department that they refuse it. Lynn was a very good example of semitone who completely ignored my advice and proved me completely wrong, so . . .Dr. Suit: Well, so did Marcia. But . . .
Dr. Urie: Yeah, but I learned that Michael was right, so I withdrew after a few years.
Dr. Suit: Well, I think that there is an increasing number of physicians who are refusing to become chief as they don't want to spend all their life worrying about money. But, thank you all very much again and very best wishes to you all.
Dr. Urie: Before we sign off I would just like to acknowledge how much work Michael put into this assignment. I feel that it's interesting and I'm pleased that I was asked to participate.
Dr. Verhey: Yes, and I am, too, very much so. It helped me revisit the good old days, some of which are not so old. It was quite a time for me. And you were a big part of it, Michael and Marcia.
Dr. Suit: Okay. Well, I see that we have come up to the 60-minute point and I would just like to express appreciation for your participation in this, Michael, and also, of course, to Marcia and Lynn.
Dr. Goitein: My thanks to all of you for being willing to give your time to this enterprise. It was a pleasure for me that we could be ‘together’ once again, albeit for so short a time, and reminisce so pleasantly together.