J Thorac Cardiovasc Surg 2007;133:1425-1427
© 2007 The American Association for Thoracic Surgery
General Thoracic Surgery
Recently, there has been a new focus on the intensity of FDG uptake in an individual tumor as a surrogate marker of the biologic behavior of that tumor. This is particularly attractive because assessment of such things as cell proliferation markers and gene expression profiling require tissue samples. Dr Downeys paper gives us insight into the true usefulness of PET and also, I believe, its limitations.
Dr Downey, I think your study corroborates several known things: the size cut-off point between T1 and T2 tumors should be lower than 3 cm; small adenocarcinomas have better prognosis, perhaps because histologies such as BAC and other GGOs are included; and most importantly, postoperative pathologic stage is still the key.
I do have one problem with the preoperative use of SUVMAX because although at your institution an SUVMAX of less than 4.3 might be predictive of good prognosis, that might not be the case at other institutions. SUV is a continuous variable, and a binary cut-off point might not be appropriate. There are many factors, as we all know, that affect the SUV determination. Therefore my first question is this: How do we use your finding of an SUVMAX of 4.3 as a prognostic cut-off point? Should we establish our own institutional cut-off point?
Dr Downey. This is a problem that has been addressed before, and it is a very important point. The standardized uptake value is not standardized between institutions. It is just standardized within an institution. Therefore a tumor measured to have an SUV of 4.3 in one institution might have a different measurement at another institution. There is interest in the nuclear medicine field for trying to come up with equipment similar to the phantom used in CT scanning that would allow standardization of SUV measurements between institutions, but thus far, there is no means of standardizing, and this just should be taken as a relative value.
The second question is about SUV being a continuous variable. I believe that there are good reasons to consider treating SUV as a binary variable. We have observed that patients who have an SUV of less than 5 almost never have nodal metastases. In our earlier article with 100 patients, there are 33 patients who had an SUV of less than 5, and only one of them had a lymph node metastasis. There are 67 patients with an SUV of greater than 5, and the incidence of nodal metastases was 33%. We have made a similar observation in patients with esophageal cancer that prognosis is best defined by dichotomizing around an SUV of 5. Therefore there might be a binary cut-off point, but this will require further research.
Dr Reed. You say that 3 clinical variables are available preoperatively to give 3 prognostic categories, yet many of us do not currently perform biopsies on small, highly PET-positive peripheral lesions before surgical intervention. You also used pathologic T size and not a CT estimate. Do you now recommend biopsy of all lesions to adequately prognosticate and perhaps, in addition, supply tissue for other markers?
Dr Downey. Again, this is a very important point. To determine the prognostic value of PET SUV, what I did was to compare PET with what would be the optimal combination of standard tests, which would be if the fine-needle aspiration preoperatively told you everything about the histology that the final pathology was going to show you, and also radiographically they were able to determine the same size that was determined by means of pathology. Despite this being the best standard that could possibly be obtained, PET still turned out to be an independent predictive variable. Obviously, there is going to be degradation from that high standard in common practice because if we try to measure things radiographically, it will only approximate pathologically measured size, and histology defined from a fine-needle aspiration is never going to be as good as the final pathology. Therefore I believe that the PET SUV might turn out to be the best independent prognostic variable that we have available in the preoperative period. I am not advocating needle biopsies in everyone.
Dr Reed. It was hoped that FDG-PET could be a useful tool in identifying patients at high risk of recurrence with each stage, particularly resected stage I and II disease. Because of your findings, would you still favor a prospective multi-institutional study looking at this issue in a homogeneous population of patients; that is, those with surgically resected stage I and II disease?
Dr Downey. There are about 10 to 12 articles that have been published looking at the prognostic significance of PET SUV. Most recently there was Dr Cerfolios article from Alabama in which he found results that were different from ours. He found that the PET did predict survival independent of pathologic staging. It was a smaller group of patients but still a very substantial number. I have not talked to him to figure out why we have the results that we have and he has results that he has. Therefore I would still consider this an open question for development. I do not know whether it requires a multi-institutional study.
Dr Reed. Finally, how would you use PET to stratify for neoadjuvant therapy? Tell us what you would recommend for a patient with clinical T2 N0 squamous cell carcinoma with an SUVMAX of 10.
Dr Downey. We have a very low threshold for recommending patients for neoadjuvant therapy; however, I think we might have identified a group of patients who might not benefit. Although it is somewhat of a philosophical question, it is not clear whether it would be worth giving induction therapy to the group with the best prognosis, who have a 3-year survival of 86%. We might be able to reduce the number of patients who get referred for induction therapy.
Dr Reed. I believe, finally, that this is a very important paper because it refines the use of PET, and I want to congratulate Dr Downey on a very nice presentation. Thank you very much.
Dr Downey. Thank you.
Dr Robert J. Cerfolio (Birmingham, Ala). Dr Miller said I can only ask one quick question, and therefore I will try to ask you 2 quick questions.
First of all, I congratulate you on your findings, and I think our findings are relatively similar, and we are corroborating one another. There are some specifics. You grouped stage II and stage III disease together. That might have been a factor, and we can talk about that later. But I think we do need a multi-institutional prospective study, and I think we should look at neoadjuvant therapy for the patients with stage IB disease. The patients with stage IB disease are the ones we all know do poorly, and I think the SUVMAX helps you identify the patients with stage IB disease who do poorly. Therefore this is my chance to try to get all the worlds experts who are here to participate in a study like that, which I think we really need to do.
The second thing is that we need to get the SUVMAX to be the same at different institutions. It is not that hard to do. The nuclear radiologists tell us they can do it. Then an SUVMAX in Brazil will be the same as an SUVMAX in New York or Birmingham.
Could you comment on the role of a prospective study looking at neoadjuvant chemotherapy for a patient with stage IB disease and how we would identify the patients with stage IB disease with integrated PET/CT and not dedicated PET?
Dr Downey. I did not mention it, but in this group we transitioned from just PET to PET/CT in the middle of this study at our own institution. I think one of the most interesting findings was the statistical phenomenon of an interaction that was seen in the patients with clinical stage IB disease, such that if it is a discriminant, it is maybe most effective in telling you who is a good- and a poor-risk patient with clinical stage IB disease. I am not sure that I would use it to start off a multi-institutional trial involving induction therapy at this point because there is no defined role for induction therapy. I think that we need to simply validate that we can identify preoperatively patients with something that correlates with pathologic stage, and that would be best done across institutions. What I think we have here is that it does show there is no relationship between PET and pathology and therefore no additional benefit of adding PET to pathology; therefore we can use pathologic stage as the gold standard, and it would be a relatively straightforward study to do to define prognostic variables preoperatively, have surgeons assign a clinical stage, and then just find out what they find at the operation a week later and see how close we can get. Then after we have done that, shown what variables work, then we can decide who should get induction therapy.
Dr Daniel L. Miller (Atlanta, Ga). Although I enjoyed your presentation, I think the most important point about this is when we look at visceral pleural invasion because we are getting to these smaller tumors2 cm, 1.5 cmand they are peripheral, and they are going to have a significant amount of visceral pleural invasion. Your greatest crossover group was stage IA to IB disease, and I think that probably was not measurement but was probably related to visceral pleural invasion. We are looking back now to our own data. There are actual lung cancers less than 3 cm with visceral pleural invasion that are actually doing worse than those of 4 cm without visceral pleural invasion. Did you look at visceral pleural invasion at all?
Dr Downey. I do have a slide that showed that, but I did not display it. The 2 areas, sort of the move from clinical to pathologic stage, if we break it down by T, N, and M, it was not in the N and the M category, it was in the T category, and it was not on measured size. The majority of them, as you pointed out and you picked up on, were going from T1 to T2 or T1 to T4. Less so, but still those are the major areas that fell off.
Dr Miller. Well, I think it is more important now with our pathologists with determining visceral pleural invasion at the time of resection because they are all getting adjuvant treatment now for stage IB disease, and if you are going to do a stage IB disease study preoperatively, we are going to miss some patients because we are not going to be able to identify that visceral pleural.
Dr Anthony P. Yim (Shatin, Hong Kong). Dr Downey, I really enjoyed your paper.
Did you look at delayed SUV values apart from the SUVMAX, and do they affect the outcome?
Dr Downey. No. Our nuclear medicine physicians have been very interested, and I have been sort of peripheral to that work, in trying to see whether there are any other parameters that work better than the SUVMAX. They have tried integrating the SUV across the tumor and various things, and they have never found anything that they think is better in any of these data sets than the SUVMAX of the primary site of disease.
Dr Yim. Just to comment, in my locality we still see a lot of inflammatory disease, such as tuberculosis, and we pay a lot of attention to the delayed value.
Dr Frank C. Detterbeck (New Haven, Conn). Rob, I have just a comment. Maybe you know the answer to this better than I do, but my understanding is that there are a lot of factors that go into SUV, and I do not think it is as simple as just standardizing the machine at one institution versus that at another. It has to do with the activity of the FDG, the timing between the injection, and when you scan. It also has to do with the size of the lesion. Once the lesion is less than 2 cm, you have to correct for the size of the lesion because otherwise you will underestimate it. It also has to do with where in the lung the lesion is because if there is a lot of movement, then you do the volume averaging on the PET scan just like you do on the CT scan. Therefore there are a lot of factors. My understanding is that determining a reproducible and reliable SUV value is not at all straightforward or simple, but perhaps you know more about this than I do or perhaps Rob Cerfolio does.
Dr Downey. I think actually Dr Cerfolio said it would be easier to standardize between institutions. I do not think it will be particularly that easy. It is the time between the injection versus when it is scanned, the dose that is given. Your point about the tumor size, though, I think is a very important one that has nothing to do with the institution. The smaller the lesions are, the more there is going to be a volume averaging. The CT scan is obtained over the length of time that the PET scan lasts, and there will be sort of a movement up and down, both smearing out the sort of SUV value over a larger volumerelatively larger volumebut also degrading the quality of the CT scan. Therefore these are all things that have to be taken into account as we work on the preoperative clinical variables.
Dr John Howington (Cincinnati, Ohio). Rob, nice presentation. There was discussion about needle biopsies and no-needle biopsies, and then the discussion led over into doing induction therapy for lesions with a high SUV. Being in Cincinnati, in the histoplasma belt, I just cringe at the thought of a patient with a solitary pulmonary nodule, with SUVs of 8, 9, and 10 that are histoplasma, being mistakenly treated for cancer. The idea of doing induction therapy in a patient without a tissue diagnosis is hard for us to swallow.
Dr Downey. I was not advocating or arguing against fine-needle aspirations. That was not the point of the paper.
Dr Cerfolio. Well, since you invited me to answer that, Frank, I would say that the formula that goes into the SUVMAX is not dependent on where the nodule is located. It is not dependent on the size of the nodule. It is a very specific formula that looks at the activity at a pixel that can add some of those variables. Actually, those are probably less than 10% differences from talking to the GE guys who do this every day. It has to do with the weight of the patient and the injected dose, and those are very easily corroborated among centers and organized. Therefore I do not think it is that hard to get an SUVMAX to be the same across the world. I will shut up now.
J. Thorac. Cardiovasc. Surg. 2007 133: 1419-1427.
|HOME||HELP||FEEDBACK||SUBSCRIPTIONS||ARCHIVE||SEARCH||TABLE OF CONTENTS|
|ANN THORAC SURG||ASIAN CARDIOVASC THORAC ANN||EUR J CARDIOTHORAC SURG|
|J THORAC CARDIOVASC SURG||ICVTS||ALL CTSNet JOURNALS|