College of American Pathologists
Printable Version

  No sure bets on micrometastases


cap today

February 2001
Cover Story

Karen Titus

Sharpen your pencils-it’s time for a pop quiz.

The subject? Better make that subjects, starting with micrometastases and sentinel lymph node biopsies and working through H&E versus cytokeratin stains; surgical false-negatives; serial sectioning; bone marrow; and, finally, breast cancer, melanoma, and, for good measure, a fistful of other carcinomas. In short, the clump of issues girding this premise: that performing sentinel lymph node biopsies and detecting micrometastases may improve prognosis as well as reduce morbidity in cancer patients.

Ready? Let’s begin.

1. Sentinel lymph node biopsy is a useful alternative to lymph node dissection for many cancers.
a) Yes b) No c) Depends d) We don’t really have a good answer for that yet.

2. Micrometastases indicate spread of disease.
a) Yes b) No c) Depends d) We don’t really have a good answer for that yet.

3. The eye-popping qualities of cytokeratin in highlighting micrometastases make it the obvious stain of choice.
a) Yes b) No c) Depends d) We don’t really have a good answer for that yet.

4. Large prospective trials will conclusively answer the above questions.
a) Yes b) No c) Depends d) We don’t really have a good answer for that yet.

Cutting to the chase, note that "d" is the correct answer to all the above, as well as to most other questions being posed in this field. Indeed, talk to even a few experts about micrometastases and sentinel lymph node biopsies, and you’ll quickly notice that words like "uncertain" and "unknown" and the phrase "yep, that’s another big debate" pop up a lot.

On the one hand, it’s a bit surprising. Physicians have known about micrometastases for years, and the literature describing them stretches back to the 1930s. Yet it’s only because of the recent development of sentinel lymph node biopsies that the finding has become so common and controversial.

"It’s now a lot easier to do a pretty intensive analysis of sentinel lymph nodes to look for micrometastases," says Donald L. Weaver, MD, associate professor of pathology at the University of Vermont College of Medicine, Burlington, and protocol pathologist for the ongoing National Surgical Adjuvant Breast and Bowel Project (NSABP) B-32 sentinel node trial.

"In the old days, even when we knew we could find micrometastases, nobody wanted to do a hundred sections on 20 lymph nodes-it would create so many glass slides you’d be buried for a week," he explains. "It was a boring old issue that no one wanted to be bothered with. But now that you can limit it to looking at one to four sentinel lymph nodes, it’s much more practical. If we find out it works well, then we can do less than an axillary dissection. So everybody is all of a sudden interested in it again. The Pandora’s box has been reopened."

"But the truth is, there’s very little scientific evidence right now," he adds. "And the scientific evidence that is available is controversial."

To say the least.

Why all the confusion? Just take a closer look at some of the issues involved.

Sentinel lymph node biopsy is a great leap forward, but it’s not always simple to do.

Because axillary dissections are associated with morbidity, "if you can just take out a couple of lymph nodes and get the same information, that would be wonderful," says Dr. Weaver.

The tradeoff for high morbidity is considerable, however. "Lymph node metastases, if they’re present, are the most ominous prognostic factor in breast cancer today," Dr. Weaver notes. Some also believe axillary dissection is a therapeutic procedure.

Mounting evidence suggests that sentinel lymph nodes yield similarly useful information in a variety of cancers. "We know that this technique in melanoma is a very good surgical technique. It’s the best possible staging technique, and it’s a very accurate predictor of outcome based on whether a [sentinel lymph] node is positive or negative," reports Alistair Cochran, MD, professor of pathology and surgery at University of California Los Angeles, and a pioneer in developing the technique for malignant melanoma in the 1980s.

The technique is arguably most advanced in melanoma. "The surgeons and nuclear medicine physicians have developed pretty good techniques for correctly identifying the sentinel nodes. The frequency of misses is very small indeed," says Dr. Cochran, who estimates that surgeons identify the right node(s) in 98 percent of melanoma cases.

He cautions, however, that concerns persist over so-called false-sentinel nodes, which can occur when lymphatic flow is blocked or deviated. "It’s a low-frequency event, but it’s one of the reasons why one has to be meticulous in choosing patients with relatively early disease. You can’t use this technique for people who are likely to have gross amounts of disease in lymph nodes."

In breast cancer, the technique is less evolved and more problematic, at least at present.

"We know there’s a false-negative rate that’s definable," says Dr. Weaver. "A group of reasonably naive surgeons would have a false-negative rate of about 10, 11 percent; when you get skilled with the technique, you can drive it down to maybe five or six percent. But the total ’herd’ false-negative rate for surgeons is probably always going to be about seven or eight percent."

The eagerness with which many surgeons are adopting the technique for breast cancer may occasionally lead to less-than-optimal surgical performance, suggests Kambiz Dowlatshahi, MD, professor of surgery, Rush-Presbyterian-St. Luke’s Medical Center, Chicago. "A lot of people are jumping on the bandwagon and doing sentinel nodes because their patients are demanding it. And they may do a lot of short-cuts." Even in the best of hands, he says, the surgical false-negative rate for breast cancer may hover around five percent.

Even highly experienced surgeons face difficulties, explains David N. Krag, MD, the S.D. Ireland Professor of Surgical Oncology at the University of Vermont and protocol PI for the NSABP B-32 study. The haphazard process by which cancer cells spread means the cells don’t always land in what would seem to be the most logical node. Similarly, the blue dye surgeons commonly use to trace sentinel lymph nodes doesn’t always target the node occupied by the migrating cancer cells. Given these and other difficulties, "We’re not going to diminish the false-negative rate down to zero-I just don’t see that happening," Dr. Krag says.

Micrometastases in melanoma offer a good model for breast cancer. And yet they don’t.

The role of sentinel lymph node micrometastases in melanoma and breast cancer may, in the end, turn out to be similar. At this point, however, the lessons learned from melanoma offer only adumbrative knowledge for breast cancer.

"The melanoma work has made it difficult for those of us in breast cancer, because of the differences in the two tumor systems," says Dr. Weaver. Yet hopes remain high. As physicians detect breast cancer earlier, more women are presenting with negative lymph nodes. "So the timing is perfect for sentinel lymph node biopsy," says Dr. Weaver. "If we find out it works well, then we can do less than an axillary dissection."

With melanoma, the goal is selective dissection of nonsentinel nodes, based on sentinel lymph node biopsy and detection of micrometastases. "If you look at 100 melanoma patients, 20 will have a positive sentinel node," says Dr. Cochran. "But of those 20, only six or seven will have tumor in the nonsentinel nodes. So at this point, we do a complete lymph node dissection on all 20 patients, because we cannot predict who is going to have a positive nonsentinel node and who is not. We’re working very hard now to develop mechanisms for predicting that."

Three factors appear to influence this difference in status, Dr. Cochran says. One is the volume of tumor that is present in the sentinel node, which is clearly related to the number of tumor cells present. The second is the volume of the primary melanoma prior to its excision. The third factor is related to the extent to which the sentinel lymph node is immune suppressed.

"Out of that trio of observations, we are developing, and will shortly submit for publication, information that should allow the prediction of nonsentinel node status," he says, adding that nonsentinel node status appears to correlate very well with eventual outcome.

"I don’t think in melanoma that anyone has any doubt that micrometastases are clinically significant-these are not innocent cells by any means," Dr. Cochran continues. The widely accepted belief is they are simply one step on a continuum that ranges from negative lymph node to single cells in the subcapsular sinus and microcolonies to more conventional macrocolonies.

While no one is exonerating micrometastases in breast cancer, neither has anyone made an ironclad case proving their guilt.

The relative ease with which pathologists can examine sentinel lymph nodes means "we can do a lot of extra work on them, and as a result we find lots of tiny little metastases, things that are one- or two-tenths of a millimeter," Dr. Weaver says. "But we don’t really know what those mean in the context of treating women with chemotherapy."

Says Dr. Krag: "There’s data to suggest they’re important, but nothing has been nailed down."

Some "ifs" to ponder: If breast cancer is a systemic disease at the time of diagnosis-which appears to be true in most cases-then the presence of a few tumor cells in the sentinel lymph node may be no more consequential than their presence elsewhere in the body. "Certainly there is empiric evidence that tumor does have access to the whole body by the time we diagnose breast cancers," Dr. Weaver observes. If the primary tumor is removed and the immune system eradicates the few remaining straggler cells, then the presence of micrometastases in lymph nodes or sentinel node would pose little added threat. But if micrometastases can evade the immune system, or if they overwhelm the immune system on reaching a certain volume or tumor burden, then their existence is more serious.

Anyway you slice it (not to mention stain it), detection methodology is the subject of extended parleys between experts.

Just how hard should you look for micrometastases?

One sliver of this controversy involves which stain to use. Screening cytokeratin-stained slides is much easier than H&E slides. "The positive cells light right up, and you can do it much more quickly," says Patrick L. Fitzgibbons, MD, chair of the CAP Surgical Pathology Committee and director of pathology at Good Samaritan Hospital, Los Angeles.

Medical oncologists tend to prefer cytokeratin stains, adds Roderick Turner, MD, pathologist at St. John’s Health Center and an adjunct member of the John Wayne Cancer Institute, both in Santa Monica, Calif. "They have more confidence when we examine specimens with keratin stains, and that’s one of the reasons we continue to use it as an adjunctive diagnostic tool. For us, it is an issue of diagnostic accuracy."

But cytokeratin stains present their own challenges, and the jury is still out as to whether findings from H&E and cytokeratin stains have comparable meaning.

A CAP solid tumor prognostic factors conference held in June 1999 (Arch Pathol Lab Med. 2000;124) danced around the issue, neither recommending nor dismissing the use of cytokeratin. "We recognized that many people were doing it, and that it was here to stay, but since there were no data from randomized trials to support its use, we thought it was improper to recommend routine cytokeratin staining," explains Dr. Fitzgibbons, who is also an advisor to the CAP Cancer Committee. "It can’t yet be considered the standard of practice." Cytokeratin staining should not be used routinely, but can be done in cases where the pathologist is suspicious but uncertain of the presence of metastases, CAP guidelines suggest.

The guidelines that emerged from the conference also called for pathologists to section nodes at 2-mm intervals. Each sentinel node should be reported separately as positive or negative, and pathologists should note the size of the metastasis.

Some, most notably Dr. Dowlatshahi of Rush, argue that even emerging standards are insufficient. In his view, pathologists need to adjust their examinations to catch up with surgical advances. With only one or two sentinel lymph nodes to examine, he says, a more extensive look is justified. He and pathologist Ken Bloom, MD, director of laboratory operations at Rush, advocate serial sections of the entire sentinel node; in their lab, a typical 1-cm-long node is sectioned at quarter-millimeter intervals, although Dr. Dowlatshahi notes the optimal interval size has yet to be determined.

The reason for the extensive cuts is simple, say the Rush physicians: A single slice reveals only whether micrometastases are present in that slice, not the entire node. Physicians may thus underestimate the number of metastases actually present.

"The analogy I’ve been using is to imagine a watermelon that has only one seed in it," Dr. Bloom says. "Let’s say we’re trying to determine whether the seed is present or not by taking a single, random slice through the melon. If we happen to hit the seed in that slice, we say the seed is there. But if we miss the seed, then we say the watermelon doesn’t contain a seed."

Such an approach "means you’re wrong a significant amount of time," he continues. "All you can really say is it wasn’t present on the cut you took."

Dr. Bloom notes that up to 30 percent of breast cancer patients recur with breast disease within 10 years, despite tumor removal, clear surgical margins, and negative lymph nodes. "So one of our hypotheses was, Are we incorrect simply because we’re not looking at the lymph node correctly?"

By stepping through sentinel lymph nodes at shorter intervals, the Rush physicians report encountering two types of metastases that would otherwise be overlooked. One is significant aggregates of tumor clearly identifiable by H&E, which are "akin to finding the watermelon seed on a different cut," Dr. Bloom explains. The second finding consists of small aggregates of cells and individual cells easily overlooked by H&E but highlighted by cytokeratin.

Fixating on the size of the largest cluster found may miss the point, Dr. Bloom says. He and his colleagues report finding showers of tumor cells that don’t form significant sizes but raise tumor burden to high levels. "Right now, if you see a met that’s bigger than 2 mm, we call that significant," he says. "But if there are three mets that are each 1 mm in size, then we don’t know what to do with those. Clearly the tumor burden is bigger than the 2-mm met, yet we ascribe significance to the larger met and not necessarily to the three smaller ones."

There’s no doubt a more careful examination of sentinel nodes will reveal more tumor cells. "But we don’t yet know the real significance of finding isolated tumor cells, whether it’s in a single section or a serial section approach," Dr. Fitzgibbons says.

"We do have to be very cautious when we report these," Dr. Bloom agrees. "We can comment that we see them, but we also have to say we don’t know their true significance."

Does the Rush approach have any merit? "Dr. Dowlatshahi has a lot of foresight-I think he’s often ahead of the curve," says Dr. Fitzgibbons. "But some of the things he’s published have not been well-received initially, and this could be one of those cases."

Dr. Dowlatshahi admits that pathologists may find the approach overly cumbersome; Dr. Fitzgibbons terms it "not impossible," although in current practice, eight cytokeratin stains for each block would be considered excessive.

Dr. Weaver flat-out says the approach is wrong with respect to guiding current therapy. Current treatments are based on historical methodology for detecting tumors; changing detection methods, he argues, desiccates longstanding predictors and could unwittingly lead to needless and potentially harmful chemotherapy.

Ironically, Dr. Dowlatshahi uses practically the same argument to arrive at a different conclusion. "This approach puts a question mark around all the past studies. So medical oncologists don’t care for what I’m saying, and they have turned a deaf ear on the subject. But I think it’s time someone blew the whistle and said, ’Wake up. There’s a serious problem in your methodology.’"

Here’s one final dose of gasoline for the fire: The current staging system (AJCC Cancer Staging Manual, 5th edition) defines a micrometastasis as being 2 mm or less. "But now the thinking is that perhaps there is a smaller size at which we don’t need to worry about them," says Dr. Fitzgibbons. Of course, no one knows what that new cut point might be, or if such a cut point even exists.

Ultimately, metastasis size may simply not prove to be critical, suggests Dr. Dowlatshahi. "Long-term, we have to switch gears and think about why some cancers metastasize and why some don’t. We should sink our effort and intellectual energy into that, rather than this business of arguing over the method of staining and number of cuts."

Do micrometastases represent a real threat? Maybe. Then again . . .

David L. Page, MD, posits that small numbers of micrometastases may be dislodged during biopsy procedures, enter the lymphatic channels, and travel passively to the sentinel lymph nodes. In comparing the existence of a handful of small tumor cells in the subcapsular sinus to the presence of metastases 2 mm or greater, he uses this analogy: "We’ve gone from the Washington Monument to a few leaves floating at the base of it."

Micrometastases that have been benignly transported "probably mean nothing," says Dr. Page, professor of pathology and epidemiology at Vanderbilt University Medical Center, Nashville, Tenn., and an advisor to the CAP Cancer Committee. Years of experience in operating on the breast and colon have demonstrated that this passive transport of cells as found in the bloodstream at time of surgical operation does not affect prognosis, he says. "Single cells and cell groups in the fluid of the lymph nodal subcapsular sinus, associated with biopsy procedures and detritus from the biopsy site, probably have nothing to do with prognosis either. It certainly has yet to be proven that they do."

With any biopsy, Dr. Krag observes, debris-including blood pigment-invariably ends up in sentinel lymph nodes. "Whether whole cells can make it that far, or whether it’s just these small molecules of pigment, no one knows. It probably does happen, but we don’t know the extent to which it happens, or what the clinical impact is," he says.

Dr. Bloom, like many others, says he doesn’t doubt that benign and malignant cells travel to sentinel lymph nodes as a result of retraction and manipulation during surgery. But in the nearly 300 sentinel node exams done using the stepped-up technique at Rush, these cells appeared in less than half the cases. "If this was something that happened routinely, we would expect to see it in virtually all of our samples," Dr. Bloom says.

Moreover, the presence of transient cells with no metastatic potential does not affect the probability of metastases existing elsewhere in the node, he says. "Finding these cells shouldn’t give you a false sense of security."

No one knows the significance of a single cluster of micrometastases either. "I don’t know if these patients really have true metastasizing cancer," Dr. Fitzgibbons says. "I wouldn’t be surprised if in some cases they turned out to have no biological significance." Yet he recognizes the difficulty pathologists have in ignoring any finding of tumor. "If you see a tumor in a node, you’re obligated to report it, and it would be very hard for most pathologists to report it as ’tumor in an axillary lymph node; however, I think it’s negative.’ Clinicians would think you’d lost your mind. Unless we can find some consistent way of separating out artifact from a previous biopsy from real micrometastases, I think we’re obligated to report them as though they are micrometastases."

It’s possible that some isolated micrometastases may simply be dormant, and never progress to form microcolonies. "With a bit of luck, they won’t wake up before you go to your just desserts," Dr. Cochran says. "The truth is, your ’cured’ cancer patient may well have colonies of dormant cells, which are held in place by the immune system or the endocrine system, or by some system that we can’t even imagine at this point."

Others caution it may be a mistake to minimize their role. Marc D. Siegel, vice president for micrometastases at Impath, reports that researchers at his company have been able to grow colonies of tumor cells from patients who have very low incidence of micrometastases. "These colonies are very predictive of outcome," he says.

Outcomes data are needed. Badly.

The next step in evaluating micrometastases in melanoma will be to determine whether their presence in sentinel nodes affects overall patient survival. An ongoing, international, multi-center trial run by Donald Morton, MD, of the John Wayne Cancer Institute, involving nearly 1,800 patients, should provide answers, says Dr. Cochran, who is director of pathology for the study.

The need for outcomes data is just as critical in breast cancer. "There’s no proof-no proof-that minimal involvement changes prognosis," says Dr. Page.

One of the more eagerly anticipated trials is the aforementioned NSABP B-32 trial. The focus of this randomized, prospective trial is to compare full axillary dissection with sentinel lymph node biopsy; a secondary objective is to ascertain whether micrometastases have an impact on survival. Begun a little more than a year ago, the study has enrolled just over a thousand patients; the goal is to enroll 4,000. Some 170 surgeons have been trained in the standardized technique for the protocol, a difficult yet necessary feat, Dr. Krag says. "We wanted to make sure the surgical variable was as small as possible, which has not been the case in other studies." Once the study hits accrual, provisional information may be available within three to four years, Dr. Weaver predicts.

Some fear outcomes data may never arrive.

"If everybody starts doing sentinel lymph node biopsies before the trials are complete, we’ll never get the answer," says Dr. Fitzgibbons. "In fact, it will prevent us from getting the answer."

That’s no idle worry. Dr. Cochran recalls that when he and his colleagues embarked on their large study comparing wide local excision with and without sentinel lymph node technique in melanoma, "We believed that people would wait until we had done the trial, and on the basis of what we found in the trial, might then begin, rather gingerly, to apply it.

"We could hardly have been more wrong," he says.

Certainly many chafe at the sluggish pace of prospective trials, and say the exigencies of cancer require action sooner rather than later.

"My colleagues say I do not have scientific evidence for treating these patients with micrometastases, but I feel convinced," says Dr. Dowlatshahi. "I’m advocating use of common sense here, rather than a hard, scientific biostatistical number."

Others express unalloyed fear that widespread adoption of sentinel node biopsy before more data become available amounts to a medical brinkmanship equally harmful to patients.

"The naysayers tell me, ’Your study is going to take 10 years, and it’s just going to come out positive anyway, and then you’ll have failed to treat all those women who could have benefited,’" Dr. Weaver says.

"But what if it’s the other way around?" he asks.

Physicians need to work through the pros and cons of doing sentinel lymph node biopsies and looking for micrometastases, Dr. Weaver cautions. "I think the initial inclination is that if you can find a tumor cell, that’s somehow of benefit to the patient, that you’ve given them additional information that will help with their treatment. But we don’t necessarily have treatments geared for this. And if we treat micrometastases the same way we’ve historically treated larger metastases, you may actually be doing patients a disservice." It may turn out that micrometastatic disease is actually some sort of minimal residual disease, for example, or that micrometastases are a marker of an early recurrent phase. "Maybe those patients are going to benefit most from alternative therapies at that point, rather than just giving them more standard therapy," he says.

Dr. Weaver acknowledges that his opponents "think I’m being archaic in my arguments. They think I’m failing to recognize there may be an outcomes association. But that’s not true at all. I myself suspect there’s going to be some link between micrometastases and outcomes. The question in my mind is, How do we treat those patients? We shouldn’t jump to conclusions before we have the facts."

"Many people have taken a few cells as an excuse to treat," adds Dr. Page. "I regard that as nothing less than the death of common sense."

And others fear that even if the data do arrive, they will be seriously flawed.

"Random biopsies that take one or two deeper cuts are a fundamental problem in studies such as the NSABP trial," says Dr. Dowlatshahi. "In the long run, I think they are going to regret using this approach." (See "detection methodology," page 34.) Indeed, he says, "most studies quoted by various experts have got serious flaws in them because they use this older methodology."

"I think the Rush approach is a great research tool," Dr. Weaver says. But with 4,000 patients in the NSABPB-32 study, "we could not justify the expense of sectioning all the way through the sentinel nodes at 0.25-mm intervals. This is why we took the strategy that we would have pathologists cut them very thin-2 mm-before embedding in the paraffin block and then perform a two-level analysis at 0.15 mm and 0.55 mm into the block with H&E and cytokeratin immunostains. This was a cost-effective strategy, but still a very expensive part of the grant," he says.

When surgeons want to do sentinel lymph node biopsies, should pathologists comply?

When he’s feeling argumentative, Dr. Weaver asks his surgical colleagues how they intend to use the information gained from sentinel lymph node biopsies.

"Some people are rational, and they say, ’I’m not going to use it for treatment. I just want to pass the information along to the patient,’" he says. "If that’s true, then finding micrometastases is much less dangerous, because the additional work you do is not going to be used to make a clinical decision."

When he’s told that the information will be used to determine treatment, Dr. Weaver has a ready answer, one that he recommends other pathologists use as well: Ask the clinicians to point to outcomes-based research that proves treating micrometastases leads to better survival. "The reason I can get away with being cocky and asking that question is because I myself have never seen any literature that’s done in a prospective manner that demonstrates a difference," he says. Most of the literature is retrospective, he says, and their data are tenuous at best. [The oft-cited Ludwig study-International (Ludwig) Breast Cancer Study Group. Lancet. 1990;335:1565-1568-doesn’t cut it, in his mind.]

M. Elizabeth Hammond, MD, says when pathologists are asked to examine sentinel lymph node specimens, any dialog with their surgical colleagues should include discussions about the surgeons’ training as well as the protocol for pathologic scrutiny. Dr. Hammond is the former chair of the CAP Cancer Committee and the organizer of the CAP solid tumor prognostic factors conference.

Dr. Krag, for his part, says he’s never done a sentinel-node-only procedure in a breast cancer patient outside of a clinical trial-but he’s quick to say he’s probably in the minority. "There are many surgeons out there who are starting to do sentinel node only, figuring it’s the standard of care. But most of us feel that’s premature."

Pressures from patients are not helping matters.

In melanoma, the general recommendation is to not perform a sentinel node biopsy for lesions less than 1 mm in thickness. "But there is no doubt, patients come in, and they ’want that new technique, doctor,’" Dr. Cochran says. Explaining the odds of their having a positive sentinel node does little to quell them. The aforementioned 20 percent figure drops to four to five percent, perhaps even less, for patients with a melanoma thickness between .75 and 1 mm; for thicknesses less than .75 mm, the possibility drops to approximately one percent. "But still, patients want it, and if they’re sufficiently persuasive, then the surgeons will generally do it."

That’s not completely bad, Dr. Cochran admits. "It is a rather small procedure, and if it gives the patient peace of mind, that’s fine."

"I must say," he adds, "that if I had the misfortune to have an invasive melanoma, I would probably want a sentinel node examined."

Patients with breast cancer are just as demanding, if not more so.

"We’re seeing a dramatic jump in the numbers of patients calling us to see if they can get micrometastases testing done," says Thomas J. Moss, MD, vice president of clinical affairs at Impath.

Dr. Bloom reports that large numbers of patients have their blocks sent to Rush for the more aggressive examination, but notes they also seek second opinions at institutions where the approach is completely different. "They make their way around the country and hear all the different opinions and assessments, from ’It’s nothing’ to ’It might be something.’ It’s confusing, to say the least."

Micrometastases in bone marrow are another wrinkle.

"Now we’re stepping off into really interesting territory," says Dr. Krag.

Exploration of this terrain was given a boost when a group of German researchers reported a year ago (Braun S, et al. N Engl J Med. 2000;342: 525-533) that the presence of cytokeratin-positive micrometastases in bone marrow increases the risk of relapse in patients with stage I, II, or III breast cancer.

The German data are striking on several accounts, not the least of which is the researchers were able to divide patients into different risk groups based on the presence or absence of micrometastases in bone marrow. "That by itself is important," Dr. Krag says.

"But what’s also important is that many of the people in the study who had cancer cells in the bone marrow are never going to have it grow," he continues. In that group of patients, the risk of dying from breast cancer wasn’t nearly as high as the incidence of bone marrow micrometastases. "Cancer cells do get loose," Dr. Krag says, "but it may or may not be meaningful in any given patient."

One of the challenges of trying to prognosticate is that research data apply to groups, not individuals, he notes. "What happens, time and again, is we’ll have two individuals with identical lymph node status, identical primary tumor status in the breast, and identical bone marrow status. And one of those patients will die of breast cancer and the other won’t."

Every step forward with micrometastases "brings us one notch closer, and we become a little more refined," he says. "But we’re still not there with individual prognostication."

On the strength of the Braun paper as well as other work by European researchers, Dr. Krag predicts a surge in research looking at bone marrow micrometastases in a host of cancers. "It’s entering into clinical trials in the U.S. more and more," he says. Though the procedure is quite painful for patients and is considered to be invasive, its complications are few, and from a surgical point of view, the technique may be less demanding than sentinel lymph node biopsies. In theory, any tumor system that metastasizes to bone would be a potential candidate.

Down the road, circulating tumor cells in blood might be another possible prognostic marker. "We’ve been looking at that in our own laboratory," reports Dr. Weaver. "But it’s proven to be a difficult challenge."

Without a clear understanding of a patient’s immunologic status, "We have a hard time knowing whether individual tumor cells in the bone marrow or blood are on their way to dying or cells on their way to creating a problem," says Dr. Hammond. The cells may circulate but not implant; they may implant but not grow; they may grow but then be rejected by the body. "It’s the same problem we face with micrometastases in the sentinel lymph nodes," she says.

Given the various twists and turns these discussions invariably take, Dr. Cochran probably comes closest to stating any sort of truth when he says, "All these questions are susceptible to study and evaluation."

Interestingly, Drs. Weaver and Dowlatshahi (along with Dr. Krag) are among the authors of a notable 1998 New England Journal of Medicine article (Krag, et al. 1998;339:941-946) looking at sentinel node in breast cancer. Certainly these two physicians are among the more visible opinion leaders in the field, and their collaboration in the NEJM study suggests they share some common ground. Yet given their differences of opinion on methodology and interpretation, it’s also apparent their disagreements, though cordially stated, cannot be easily glossed over.

When asked about this, Dr. Dowlatshahi laughs and says, "We haven’t drawn daggers." But like so many in this field, these two find themselves on the same page-and miles apart.

Karen Titus is a CAP TODAY contributing editor and co-managing editor.