December 2011
Feature Story

William Check, PhD

At the scientific plenary at CAP ’11, titled “Sentinel Lymph Node Dissection for Melanoma: Is the Standard of Care Ahead of the Evidence?,” the speakers staked out divergent positions, even using different approaches in their arguments. J. Meirion Thomas, MS, FRCP, FRCS, answered the question in the affirmative. He contended that the widespread use of sentinel lymph node biopsy followed by immediate dissection of regional lymph nodes in patients with tumor in the sentinel nodes “is well ahead of published evidence and should not be done.

“Sentinel lymph node biopsy is a prognostic tool,” Dr. Thomas, professor of surgical oncology at Imperial College London and the Royal Marsden Hospital, said. “That is where the story should stop.” He presented statistical and methodological re-analysis of the clinical trial called MSLT-1 on which the use of SLN biopsy with completion lymphadenectomy is based. “There is no evidence that it is of any other benefit [than prognosis],” Dr. Thomas said. “Doing this procedure amounts to excessive and unwarranted surgery. Many patients are wrongly upstaged and wrongly informed about the benefits of sentinel lymph node biopsy.”

He called for medical professionals to be “much more honest about the benefits and complications” of the procedure, and said, “I don’t believe that it should be done at all. I don’t do it.”

Speaking on behalf of SLN biopsy followed by immediate removal of the draining lymphatic basin when tumor is found, Vernon K. Sondak, MD, too, cited data from MSLT-1. “The bottom line is that we have a lot of data and experience about sentinel lymph node biopsy. Evidence from the RCT [randomized controlled trial] shows that melanoma patients who had positive sentinel lymph nodes and early lymph node surgery actually lived longer without evidence of recurrence,” said Dr. Sondak, chair of the Department of Cutaneous Oncology and a surgical oncologist at Moffitt Cancer Center, Tampa, and professor of surgery at the University of South Florida. On the other hand, Dr. Sondak said, “We all agree that sentinel lymph node biopsy is not designed to cure anybody by itself. You don’t live longer if you have this procedure than if you don’t.”

However, Dr. Sondak’s strongest appeal to the audience of pathologists focused on the individual patient. “If I were a patient with melanoma in a lymph node,” he hypothesized, “I would want to get it taken care of now. Even if absolute survival is no better, I would rather have it out. Most of my patients come to the same conclusion.”

Luke A. Perkocha, MD, MBA, suggested SLN dissection for melanoma as a plenary topic and assembled the panel. “While there is no doubt that the results of sentinel lymph node biopsy have prognostic value in select populations of melanoma patients, its usefulness [to guide treatment] in an individual patient remains more controversial,” Dr. Perkocha told CAP TODAY. Many in the audience may not have known there was controversy, says Dr. Perkocha, associate professor of pathology and dermatology and associate director of surgical pathology at the University of California, San Francisco. “Contrary to what some people say or how often the procedure is presented as the standard of care, it is not settled.”

He considers it a timely message, given the Task Force on Preventive Services’ recent announcement that it recommends against PSA testing. “It illustrates that it is not uncommon in medicine for things prematurely to become the ‘standard of care’ and later to be proven to be useless, or in some cases harmful,” Dr. Perkocha says. He cites irradiation of children with an enlarged thymus, once commonly done, as another of many examples. “Pathologists as true medical scientists may be in a position to understand this more than some other physicians.”

Speaking with CAP TODAY, Boris C. Bastian, MD, who moderated the plenary and gave an introductory talk, commented on the important distinction between a practice that is widespread and one that is standard of care. “‘Standard of care’ does not mean common practice,” Dr. Bastian, professor of dermatology and pathology and the Gerson and Barbara Bass Bakar distinguished professor of cancer biology at UCSF, said. “It means there is conclusive evidence that the benefits of a procedure outweigh its risks.” It is true that surgeons worldwide commonly perform SLN biopsy/completion lymphadenectomy. In Dr. Bastian’s view, however, this does not make it the standard of care.

At the ASCP annual meeting in 2010, Mark R. Wick, MD, professor of pathology and associate director of the Division of Surgical Pathology and Cytopathology at the University of Virginia Health System, gave the Arthur Purdy Stout Society lecture on SLN biopsy/completion lymphadenectomy. Summarizing his message in a recent CAP TODAY interview, Dr. Wick said, “Sentinel lymph node biopsy is only a staging procedure. This is true whether we are talking about melanoma or breast cancer or lung cancer.”

According to Dr. Wick, SLN biopsy/completion lymphadenectomy is based on old concepts of metastasis that came from William Halsted, the first chairman of surgery at Johns Hopkins University, in the late 1800s. “Halsted postulated that metastasis is linear and goes from one structure to another sequentially,” Dr. Wick said. “He introduced removal of the primary tumor plus draining lymph nodes.” Halsted believed that the Halsted radical mastectomy would interrupt the ability of breast tumors to involve visceral organs. However, it was demonstrated in a series of trials in the 1970s and 1980s to provide no additional benefit for most breast cancer patients.

“Even though this concept is now over 100 years old, many physicians still cling to it,” Dr. Wick said. “For some time we have had data on the biology of metastasis that gives the lie to the notion of serial spread. A malignant tumor that can metastasize will exercise that ability right away and will metastasize to lymph nodes and visceral organs simultaneously.” Dr. Wick has published these arguments (Wick R. Pathology Case Reviews. 2008;13:1–7).

To begin his introductory talk at the CAP ’11 plenary, Dr. Bastian, too, noted that the idea that removing regional disease would prevent spread to distant sites is based on the biological concept of stepwise progression. “Is the difference between stage three and stage four a biological truth,” he asked, “or an artifact of our methods of detection?” The alternative, he said, is time difference—disease spreads widely from the outset but shows up later at distant sites because it does not grow at the same speed everywhere. “The argument is made that melanoma is different,” Dr. Bastian tells CAP TODAY. “But it isn’t. In fact, melanoma is probably a tumor where systemic spread happens earliest.”

He then turned in the plenary to the critical study that provides the most robust evidence to assess the utility of SLN biopsy in melanoma, MSLT-1 (Morton DL, et al. N Engl J Med. 2006;355:1307–1317). This multicenter randomized controlled trial enrolled 1,269 melanoma patients with 1.2- to 3.5-mm thick (“intermediate”) lesions. All had their primary tumor removed, while one group also had SLN biopsy. In the biopsy group, patients whose SLN biopsy showed tumor had immediate removal of all nodes in the draining basin. In the observation arm, removal of regional nodes was done only when palpable disease was detected in the nodal basin.

Two clear-cut, uncontroversial conclusions issued from this study, Dr. Bastian said, along with three controversial ones. The clear-cut:

• There was no significant difference in overall survival between the two arms.
• Within the biopsy arm, patients with negative SLN biopsy fared better in overall survival.

The disputed interpretations are as follows:

1. Compared with observation, biopsy can identify patients who can benefit from immediate completion lymphadenectomy.
2. SLN biopsy/completion lymphadenectomy prolongs disease-free survival.
3. SLN biopsy/completion lymphadenectomy provides prognostic information independent of histopathological and demographic information.

Conclusion No. 1 rests on the finding that patients in the biopsy arm who had metastases survived longer than corresponding patients in the observation arm. Dr. Bastian said this conclusion is disputable because it comes from a posthoc subgroup analysis, which further assumes that patients with disease in the two groups are identical. However, the analysis actually compared the survival of patients in the biopsy arm who had a positive biopsy plus those who had later lymphatic disease to those in the observation arm who had clinically detected lymphatic disease. All patients in the observation arm, therefore, had a sizable metastasis, Dr. Bastian noted, whereas some of those in the biopsy arm had micromets, just a few cells, picked up on SLN biopsy.

Conclusion No. 2 rests on a posthoc comparison of the incidence of relapses in the regional basin between the two arms. Again, Dr. Bastian said, the two groups were not identical: Some of the patients in the biopsy group already had their regional lymphatics removed, so they could not get regional metastases. “This comparison skewed the results in favor of sentinel lymph node biopsy and made it look as though those patients were living longer without disease,” Dr. Bastian said. “What should have been compared was the time interval to distant disease.”

Conclusion No. 3 enters traditional pathology territory. Using data from the study, Dr. Bastian calculated that the sensitivity of SLN biopsy was 33 percent and specificity was 86.6 percent, for a positive predictive value of 26.1 percent and negative predictive value of 83.3 percent. “This means that 17 percent of patients in whom biopsy is negative will die of melanoma,” Dr. Bastian said. Conversely, only 26 percent of patients who have a positive SLN biopsy will die of their disease. “No lab test would be introduced into clinical use with these metrics,” Dr. Bastian emphasized. Nor is it likely that the Food and Drug Administration would approve a test with these numbers as an indication to put a patient on biological therapy, for example. Though this is a surgical procedure, Dr. Perkocha told CAP TODAY, the results are being used in ways similar to those of a laboratory test. “It is thus not unreasonable to impose the same metrics,” Dr. Perkocha says.

At the outset of his talk, Dr. Thomas said medical practice should be based on evidence. “Standard of care has no definition in medicine,” he said. “The concept is based in law and is an essential component of medical malpractice. In medicine it can be self-awarded by a group of like-minded individuals to provide impact and authenticity to a point of view.”

With regard to SLN biopsy in melanoma, Dr. Thomas said, “The only thing we know for sure is that it is a staging and prognostic tool. It is one of many prognostic factors. Every other advantage is wishful thinking and cannot be confirmed statistically.”

As Dr. Bastian had foreshadowed, Dr. Thomas focused his analysis of the statistical accuracy of MSLT-1 on two claims: that it demonstrated a disease-free survival advantage in the biopsy arm and that a subgroup analysis showed a 20 percent survival advantage for early (immediate) lymph-adenectomy. “Both of these conclusions are wrong,” he said unequivocally.

“The apparent disease-free survival advantage is a result of trial design and does not imply a therapeutic benefit for the patient,” Dr. Thomas said. Disease-free survival is the time from wide excision to recurrence at any site. For the patients in the trial, the regional nodes were the most likely site. “However,” Dr. Thomas said, “patients in the biopsy arm with positive sentinel nodes had lymphadenectomy. So the patients most likely to develop palpable nodal recurrence were identified and had prophylactic surgery. Inevitably there would be more nodal recurrences as site of first recurrence in the observation arm.”

Dr. Thomas supported his contention by comparing sites of first recurrence in the two arms. Two categories differed. Palpable tumor in the regional nodes was found in 13 percent of patients in the observation arm compared with 4.2 percent in the biopsy arm. Distant recurrence, on the other hand, was actually less in the observation arm, 7.8 percent, than in the biopsy arm, 11 percent. So the difference in nodal recurrence accounted entirely for the difference in total recurrence (26.8 percent versus 20.7 percent).

“The only endpoint that matters is distant recurrence,” Dr. Thomas said. He made this point in a letter to the National Cancer Institute, which funded the MSLT-1 trial. On June 11, 2007, he received a reply agreeing with his analysis. “[C]alculation of either distant disease-free survival or excluding nodal recurrence from the calculation is appropriate,” the letter said. The NCI advised the study’s first author, Dr. Morton, to calculate disease-free survival without nodal recurrence at the next interim analysis. “In 2006 he was instructed to do that,” Dr. Thomas said in a CAP TODAY interview, “and he has still not done that.”

To counter the study claim of a 20 percent survival advantage at five years, Dr. Thomas noted that this rests on a posthoc analysis between unequal groups, as Dr. Bastian had described.

In addition, Dr. Thomas noted that SLN biopsy has a substantial false-positive rate. This is because it counts all micrometastases. He called it “biologically insane” to assume that all micrometastases will progress to palpable nodal disease. Based on the initial report of MSLT-1, he calculated the false-positive rate as 24 percent.

“We are told that patients are the drivers of sentinel lymph node biopsy because they wish to know their prognosis in the hope that they are node negative,” Dr. Thomas said. “But are patients fully informed?” He wondered whether they are told that 15 percent to 20 percent of node-negative patients will have a recurrence by five years. “Are they told that the incidence of false-negative results is 12.5 percent, as was published recently?” he asked. “Or that the complication rate of sentinel node biopsy in MSLT-1 was 10 percent to 26 percent?”

In an interview, Dr. Thomas expanded on these thoughts. “Patients with melanoma are very frightened,” he said. “Unlike colon or breast cancer, they know there is no systemic or adjuvant therapy that can help them. A surgeon says ‘We have a procedure that offers you a survival advantage. And if your sentinel lymph node biopsy is positive, we can offer you more advantage.’ Suddenly it’s a different disease. Someone has offered you hope. Whereas in fact there is no statistically valid evidence that it provides any advantage in disease-free or overall survival.”

Dr. Sondak began his talk with a strong personalized appeal to the audience. “Imagine for a moment that you are the patient with a thin, an intermediate, or a thick melanoma,” he said. “And ask yourself if, after you hear all of the available data, do I want to know this information? Will it have value to me personally? Because individual patients, that is who we are counseling. That is who we are going to try to make the decision for with the data we have today.”

He then listed several reasons why SLN biopsy is standard of care.

First, it is a reliable prognostic factor, a fact with which the other speakers had agreed. Again, Dr. Sondak presented a strong personal appeal to the pathologists. “Consider a case with a Clark level 5, 4.5-mm melanoma,” he said. “None of you would want one of these on your body. Does that mean that every patient like this will die of this disease? Does it mean that it is prognostically false-positive to sign that out as thick melanoma and not melanoma in situ? Did you commit malpractice because you led the surgeon to conduct ‘excessive surgery’ on that patient when the report comes back as no residual melanoma? None of those things are the case. We do the best we can with the information we have that is reliable, not infallible.”

Dr. Sondak’s second reason was that there is no available imaging technique that reliably detects small nodal deposits or micrometastases that are routinely found on SLN biopsy.

Third, he said, “Disease detected on sentinel lymph node biopsy is clinically significant and predicts a greater number of recurrences.” Dr. Thomas had argued that some of the micrometastases detected on SLN biopsy were clinically insignificant, based on the fact that the number of recurrences in the observation arm remained less than in the biopsy arm at five years. Study author Dr. Morton has contended that there would be a “surge” of cases in the observation arm by eight years to close the gap. As Dr. Sondak put it in an interview, “Sentinel lymph node micromets grow up to be macromets.” In his talk, Dr. Sondak took this argument further.

“I believe that at 20 years the number of palpable lymph nodes discovered [in the observation arm] will be higher than [in the biopsy arm].” His reasoning: “Occasionally pathologists called a lymph node negative, but there were a few micro cells in the lymph nodes that you missed. Not because you didn’t see them, but because the cutting instrument didn’t cut through that LN. If left for 15 or 20 years, some of those will develop into palpable metastases and the number of node-negative patients will be fewer in the observation arm.”

Fourth, SLN biopsy allows earlier institution of lymphadenectomy and adjuvant interferon therapy. In addition, Dr. Sondak said, “The risk of lymphedema is less and it is less severe when I do a node dissection after a sentinel node biopsy than when I wait for a delayed nodal recurrence.”

Fifth, “I believe it is probable that there is a therapeutic impact for node-positive patients. While we wait for more data I ask you to put yourself in the patient’s place. Ask yourself, Would I roll the dice today or would I wait for the next analysis of the trial?”

Referring to the claim in the original publication of a survival advantage for early lymphadenectomy, Dr. Sondak said, “This is a posthoc analysis. There is no two ways about it. It’s a limited analysis; it’s not an invalid analysis. The data is the data.”

Dr. Sondak’s final reason: “Our patients really want to know. You would want to know if your sentinel lymph node was positive if you had a thin, intermediate, or thick melanoma,” he said. “It’s the best information we could give you today about the state of your disease and would allow you to make the most informed choices.”

In an interview with CAP TODAY, Dr. Bastian called this last point “the argument of last resort by surgeons.

“Even if all those metrics of the clinical trial really don’t show tangible benefit, patients just want to know. Basically their argument is very difficult to refute without data or questioning patients. What patients want to know is whether they will die of disease. Patients are desperate for information. It is absolutely true that most patients will want to know [their SLN status].” However, Dr. Bastian said, “The information that sentinel lymph node biopsy provides is presented as if it were answering that question about life or death. It does that only to a very limited degree. The fact is that [surgery based on SLN biopsy] doesn’t improve overall survival. That proves that tumor is present not only in the sentinel node but in many sites in the body. And removing tumor cells in the lymph nodes wouldn’t help other sites.”

Use of SLN biopsy followed by axillary dissection has been extensively explored in breast cancer. According to Dr. Wick, studies in the 1970s and 1980s disproved the idea that positive regional nodes are the source of serial distant disease. “[The data] said quite clearly that breast cancer metastasis is not linear,” Dr. Wick says. “All that a positive lymph node in the axilla tells you is that the patient has systemic disease.” A summary of one series of trials concluded that “positive regional LNs are indicators of a host-tumor relationship that permits the development of metastases; they are not important sources of distant disease” (Fisher B, et al. Cancer. 1981;48:1863–1872).

More recently, a randomized controlled trial in women with invasive breast cancer and SLN metastasis found that axillary dissection had no impact on survival (Giuliano AE, et al. JAMA. 2011;305:569–575). A companion paper asking whether this procedure is “becoming obsolete” found a “diminishing rate” (Weber WP, et al. Ann Surg Oncol. 2011; Jun 7. [Epub ahead of print]). “In breast cancer I think the field has moved to the ‘been there, done that’ stage,” Dr. Bastian says.

He cites a recent retrospective study in SLN biopsy-positive melanoma patients of those who had lymphadenectomy and those who did not. “Recurrence-free survival and disease-specific survival were similar between the groups,” the authors found (Kingham TP, et al. Ann Surg Oncol. 2010;17:514–520). “It remains unclear whether [lymphadenectomy] must be performed in all melanoma patients with a positive SLN,” they concluded.

Dr. Thomas finds the claim of lower incidence of lymphedema after immediate lymphadenectomy, 12 percent, than after delayed lymphadenectomy, 24 percent, to be flawed. “Patients with early lymphadenectomy had biopsy,” he points out, “and sentinel node biopsy in itself can cause lymphedema.” Lymphedema from the SLN biopsy must be added to give a true incidence for this strategy.

Dr. Sondak disagrees with Dr. Thomas’ interpretation of these data. “After several years fewer people showed up with a lump in their lymph node than were told they had positive nodes based on sentinel node biopsy,” he says. “Does that mean there is a group of patients in whom that is a micro finding and not of clinical significance? Are there patients in whom it would not have progressed?” His answer—“No. We reject that there is cancer that has no clinical or biological significance. All cancer is a bad thing and you take it out.

“We can’t say at six years there is an obvious benefit,” Dr. Sondak concedes. On the other hand, he says, “We have to recognize that things might change over the years. But our patients don’t have six years to wait. They have to make a decision based on the best available data we have today.”

Of all medical practitioners, Dr. Sondak thought pathologists would be the ones to understand the “hazards of predicting the future.”

“When a pathologist looks at one of my patient’s melanoma and says it looks really bad, it has a lot of poor prognostic features—a large, very thick lesion with deep extension, ulceration, high mitotic rate—that pathologist is communicating to me and to the patient that we should be very worried. But many people with high-risk, poor-prognosis lesions many years later are still walking around cured of their melanoma just by removing it.” About such patients, Dr. Sondak asks, “Did the pathologist lie to me? Give me false information? Needlessly alarm the patient? No, they called it as they saw it.” In such cases, Dr. Sondak says, “We never say that because this patient’s melanoma didn’t come back that means the initial report was wrong. No, the report was right. The fact that the patient was more likely to die of melanoma was true. But that doesn’t mean they were certain to die of melanoma.

“This whole mathematical argument is irrelevant to an individual patient,” Dr. Sondak concludes. “Patients don’t care what is true for the whole species of Homo sapiens. They just want to know, What does this mean to me?”