April 2011
Feature Story

Karen Titus

Unlike certain things—love, say, or weapons of mass destruction—micro­metastases aren’t incredibly difficult to find. By carefully cutting the paraffin tissue blocks of sentinel lymph nodes and using immunohistochemical staining, pathologists have an easy task, relatively speaking.

But then what? Or, to put it another way—so what?

Those questions have nagged pathologists and clinicians for years, even as the former dutifully reported to the latter their findings of micrometastases and isolated tumor cell clusters. “The pathology reports come out with a bunch of little letters describing those smaller metastases, but no one really knows what it means,” says David Krag, MD, the S.D. Ireland professor of surgical oncology, University of Vermont, Burlington. Does their presence or absence bestow prognostic value? Can they help physicians decide treatment? What’s the best way to look for them? Now, with the recent publication of results from the National Surgical Adjuvant Breast and Bowel Project trial B-32, answers are at long last forthcoming.

The answers are important, and perhaps not entirely unexpected. Over the years opinions of the usefulness of finding smaller metastases have flown back and forth like snowboarders in the halfpipe. Most recently, small studies—and growing opinions—have suggested they’re really not all that valuable clinically. The results of B-32 (Weaver DL, et al. N Engl J Med. 2011;364:412–421) appear to confirm that belief.

Not too long ago, however, the reigning philosophy leaned in a different direction: Micrometastases can be found; ergo, they must be important, says Donald Weaver, MD, protocol pathologist for the B-32 trial. Exactly how they might be important was never all that clear, but the simplicity of the argument was appealing: The sooner cancer is detected the better.

Traditionally, pathologists have relied on categorizing nodes as either positive or negative, says Dr. Weaver, who is also professor of pathology, University of Vermont College of Medicine, Burlington, and attending in pathology and director of surgical pathology, Fletcher Allen Health Care. With the appearance of the AJCC Cancer Staging Manual’s 6th edition (a 7th took effect Jan. 1, 2010), physicians began to recognize the importance of disease volume.

But habits cling like kudzu. “Because of that historical mindset, detection of metastases seemed to take precedence,” Dr. Weaver says. “It’s also counterintuitive for people to think that something you find might not be significant.” He hopes the B-32 results will inject scientific sense back into the discussion.

For Dr. Weaver, the study drives home two important findings.

First, he says, is the proof, finally, of the hypothesis that nodal tumor burden is a continuous variable. “I’ve certainly spent an awful lot of my time trying to argue to people that it’s nodal tumor burden that’s important, not just whether it’s positive or negative,” he says.

Second is that small metastases—micrometastases (>0.2 mm and 2.0 mm) and ITCs (<0.2 mm in the greatest dimension)—are not good predictors of recurrence in the context of a highly treated population.

The publication of the findings “almost feels anticlimactic,” Dr. Weaver says, “because it’s been an issue for so long.” The study began more than a decade ago, and it took longer than expected to obtain enough recurrence events (50 percent) to spark the statistical analysis.

In the trial, 5,611 women with breast cancer were randomly assigned to sentinel-lymph node biopsy with axillary dissection or SLN alone; ultimately, pathological material was available from 3,887 patients (followup information was available for all but three of this latter group). At participating sites, pathologists used H&E staining to examine paraffin-embedded tissue blocks cut at 2-mm intervals, which would allow for detection of macrometastases (>2.0 mm). Routine use of IHC was prohibited, though it could be used to confirm/refute findings suggestive of metastases on initial sections. In followup exams, Dr. Weaver and his colleagues used IHC to look for occult metastases, using additional cuts 0.5 mm and 1.0 mm deeper in the blocks. The treating physicians were not told of the findings; rather, treatment was based on primary tumor characteristics and the original pathology report of negative nodes.

The researchers detected occult metastases in 15.9 percent of the patients (11.1 percent ITCs, 4.4 percent micrometastases, 0.4 percent macrometastases). These metastases proved to be an independent prognostic variable in patients whose lymph nodes appeared to be negative on initial examination. Yet the magnitude of difference in overall survival at five years was only 1.2 percentage points. The difference in disease-free survival as well as in distant-disease-free interval was 2.8 percent.

“These are really close to meaningless numbers because they’re so small,” says Dr. Krag, a paper co-author. The study is “unambiguously coming out with a statement that there’s no clinically meaningful survival difference. It doesn’t justify changing practice across the board.”

But, says Dr. Weaver, it does prove the concept that increasing volumes of nodal tumor burden is bad, even at the low end of the spectrum, i.e. the ITCs and micrometastases. “We’ve now got data that supports the continuous variable on which we have imposed artificial cut points,” he says. “It’s just a nice piece of science data to establish.”

Dr. Weaver says he was happy that among the 3,887 cases, the researchers found only 14 macrometastases on closer examination. “The most important thing we did with the pathologic evaluation of the sentinel nodes is cutting nodes at 2-mm intervals,” he says. “If you just do this protocol [H&E staining, 2-mm cuts], you’re not missing disease—the larger metastases—that has been shown over and over and over again to matter.”

Patients with macrometastases or a micrometastasis had a four percent risk of dying from breast cancer, while those with an occult ITC had a 2.2 percent risk. There are those who will find these numbers significant for treatment purposes. “Some people say, ‘I’m a one percent person,’” says Dr. Krag. “If they want to go ahead and do immuno, it’s OK with me. But across the board, I know it’s an extremely small impact.”

Dr. Weaver says he questions the value of acting on these small percentages, given that most women are highly treated. “When we were doing the analysis, I started to ask the question, Are these things predictive?” Dr. Weaver says. They weren’t. “They were terrible at predicting,” he says.

Of the 616 women in whom occult metastases were detected, 80 to 81 percent were free of disease. By making the bold assumption that women would benefit from added treatment that they’d otherwise not get, he says, “You’d be overtreating about 80 percent of the women.”

This reinforces, in Dr. Weaver’s mind, the idea that occult metastases are a relatively minor variable in the grand, primary tumor scheme of things. And this, he says, is good news for women. “You don’t have to be too worried about whether someone’s done a hundred sections on your lymph node as long as they did a good job of slicing it up on the gross bench, and did one section from the top of the tumor.”

The B-32 trial has already been causing ripples among pathologists and clinicians. Not surprisingly, so has the American College of Surgeons Oncology Group Z0011 trial, which cannonballed into the pool—or, to be more precise, JAMA—a week later (Giuliano AE, et al. 2011;305:569–575) with its suggestion that some women with invasive breast cancer might not need axillary lymph node dissection.

Syed Hoda, MB, BS, professor of pathology, Weill Cornell Medical College, New York City, conducted an informal survey—“not scientific by any stretch of the imagination,” he says—of some 50 U.S. pathologists, after he had been asked his opinion of the two studies by the New York Metropolitan Breast Cancer Group (a multispecialty group of surgeons, oncologists, radiation therapists, and pathologists).

The 30 responders (with some changing their opinions after submitting their initial answers) were more or less equally divided among those who advocated continued use of cytokeratin in SLN (this group had a slight majority); those who never used cytokeratin; and those who are thinking about discontinuing cytokeratin, at least in some cases.

One respondent in the first group wrote, “I think getting to the diagnosis faster and with greater ac­curacy is in the best interest of both the pathologist and the patient.” Another wrote, “Cannot imagine stopping use of CK on SLN.”

Weighing in for the second group, one pathologist responded, “I am pleased to say that we have never done routine CK stains on sentinel lymph nodes.... Given the recent data from the trials you cited, I feel quite vindicated that we took the correct approach for all these years. I honestly think that the major reason most pathologists jumped on the CK IHC bandwagon so early and so eagerly is 1) that it takes far less time for a pathologist to look at CK IHC stains of lymph nodes than it does to look at H and E sections and 2) they can bill for it.”

One wag declared, “Cytokeratin on sentinels is for wimps!”

And perhaps summing it up best, another pathologist wrote: “I think this is a difficult question.”

Indeed, as Dr. Hoda points out, two of the largest academic institutions with highly active breast services, Memorial Sloan-Kettering Cancer Center, New York, and the University of Texas M.D. Anderson Cancer Center, Houston, have taken opposite tacks. One recently did away with deeper cuts and cyto­keratin altogether, while the other continues to use cytokeratin. “Both approaches have merit, and each viewpoint has to be taken seriously,” he says.

Dr. Hoda presented his findings at a recent meeting of the New York Metropolitan Breast Cancer Group, where a surgeon and an oncologist also spoke. All three, he says, tried to give objective views. They then polled the audience—a multispecialty group of about 100—about whether anyone had changed his or her mind. “Only a few hands shot up,” he says. It may take 10-year (or longer) followup data to convince physicians, he says. (Dr. Weaver notes that the B-32 data will be revisited at the 10-year mark.)

In New York City, even patients take an active part in the IHC versus H&E debate, says Dr. Hoda. “New Yorkers are very informed.” This may be one reason institutions that do IHC may be unwilling to drop the hunt for smaller metastases.

And hospitals themselves might be performing a bit of one-upmanship, he suggests, with some physicians telling patients that IHC represents a higher level of care. “In New York there are 20 or so great institutions within 30 minutes driving distance of each other,” says Dr. Hoda; New York-Presbyterian Hospital (where Dr. Hoda is attending pathologist), for example, is right across the street from Memorial Sloan-Kettering. “Twenty paces,” as he puts it. “It doesn’t take much for the patients to go to another place.”

For the record, Dr. Hoda and his colleagues plan to keep using IHC and deeper levels. In a busy breast cancer service, he says, cytokeratin boosts efficiency. “And it makes us accurate,” he says, noting there are certainly cases in which subtle—and even not so subtle—metastases can be missed. In particular, he says, metastatic lobular carcinoma, which can mimic a normal lymph node, can be easily overlooked on H&E.

Dr. Weaver, not surprisingly, has some vivid opinions about IHC and deeper cuts. For years, he’s played the part of mild-mannered prophet, warning that the advent of SLN biopsies put undue emphasis on finding the tiniest metastases. “It got incorporated into practice as almost the de facto standard without evidence to suggest it was valuable for patients,” he says, firmly but nicely.

He predicts the B-32 results might help others sever their ties to deeper cuts, if not IHC. “My guess is that most pathologists will be happy enough getting rid of the deepers,” he says. “But you’re going to see a real schism between using and not using IHC”—which seems to be borne out in Dr. Hoda’s survey.

The most defensible argument for using IHC is that it allows pathologists to screen slides more quickly, he says. But that needs to be balanced out by the anxiety patients and physicians could feel when micromets and ITCs turn up. (Dr. Weaver says he doesn’t suffer from the opposite anxiety—that he’ll miss something using H&E. “I’ve said to people publicly, when they ask whether I’m worried about missing something, ‘No, I’m not. If I miss a few cells, what’s the big deal? We know we haven’t harmed the patient.’”)

Dr. Weaver takes a dim view of those who might view IHC solely through the economic lens. “You get an extra charge that you can layer onto every sentinel node submitted by the surgeon—even Medicare will pay it,” he says. Such thinking, he says, is “not altruistic at all,” though he acknowledges it’s never easy to give up a potential revenue source.

Pathologists still need to report findings of micrometastases and ITCs, of course, but the pressure should be off their shoulders to conduct a top-to-bottom, heave-every-bit-of-hametz search for the tiniest of cancer cells.

Dr. Weaver shares the story of an American College of Surgeons meeting a year ago, where he was the last in a series of speakers on SLNs. The speaker before him, Clifford Hudis, MD, chief of the breast cancer medicine service at Memorial Sloan-Kettering, told the audience—“It was a room of 2,000 surgeons, or something,” Dr. Weaver says—“Sometimes I don’t even need to know the node status.”

“He kind of took my thunder away,” says Dr. Weaver. “But sometimes it’s true. The world has changed.”

If there’s any doubt about the rapid flips in breast cancer discussions, one need look no further than the JAMA paper, which quickly cartwheeled its way into the lay press. (“Lymph node study shakes pillar of breast cancer care,” The New York Times announced.)

The so-called Z11 trial closed early for lack of patients. Patients with node positivity didn’t want to be randomized to the “no further surgery” arm of the trial, and surgeons felt it would be unethical to withhold the more extensive surgery from patients.

Now comes the foreshortened study, suggesting, ironically, that less surgery might be better, at least for certain patients meeting specific criteria.

Patients may be convinced, but despite the excitement generated by the JAMA paper, not everyone is jumping on the trampoline. As Dr. Weaver cautiously puts it, “I’m not sure I have the paper all figured out yet.”

One fact in particular gnaws at him: the study’s lack of power. The study was probably ahead of its time, he says—remember, it was launched in the more-is-better era—so it accrued only 891 of the 1,900 patients researchers hoped for. But if it holds up over time, it would be one more reason to base treatment on primary tumor characteristics.

Again, a little historical context might be helpful. Remember, says Dr. Weaver, that breast cancer patients, by and large, have clinically negative nodes—which is far from the case decades ago, when virtually every patient presented with palpable nodes and a 75 percent risk of having pathologically positive nodes. The advent of widespread screening “turned that whole thing upside down. So many women are screened out before they even get the sentinel node biopsy,” says Dr. Weaver. At Fletcher Allen, he estimates, only 15 percent of the women who undergo SLNs have positive nodes.

Eventually, he says, SLNs could migrate to more of a niche procedure, done only in a small group of women; the rest might reasonably rely on axillary ultrasound and fine needle aspirations of suspicious nodes. SLNs are likely unnecessary in the triple-negative tumors, since they will be treated with high-dose chemotherapy regardless of node status. Nor are they needed prior to neoadjuvant therapy. Removing a node makes it impossible to assess whether the patient has had a complete pathologic response. Better, says Dr. Weaver, to do an FNA of a node, then re-evaluate with a sentinel node biopsy post-therapy.

HER2-positive tumors are more of a conundrum. Aggressive treatments may reduce the need for SLN biopsies. On the other hand, a negative SLN can be a favorable prognostic factor.

Dr. Krag is also withholding judgment of the JAMA paper. “I can’t really answer what this means yet. The numbers look good, but they’re very underpowered,” he says. “I really want to understand this thing better.”

“It’s just one study,” agrees Regina Rosenthal, MD, a breast surgeon in private practice in Salt Lake City. “And it closed early. But it’s probably going to turn our practices upside down.” Already patients who are familiar with the article have told her they don’t want her to do an axillary dissection, regardless of their lymph node status. Clearly, statistical concerns are not uppermost in their minds.

Prior to her interview with CAP TODAY, Dr. Rosenthal says, her first surgical case of the day involved a patient with a localized tumor who was scheduled for a lumpectomy and a SLN biopsy, followed by an axillary dissection if the sentinel node was positive. The day before, she and the patient discussed the JAMA paper, with the patient saying she didn’t want an axillary dissection even with a positive lymph node.

“I grilled her again right before surgery,” says Dr. Rosenthal; the patient remained adamant. When the pathologist entered the OR, Dr. Rosenthal told him of the patient’s wishes.

The first sentinel node was positive.

The second sentinel node was negative.

Dr. Rosenthal did not perform the axillary dissection. “I felt like I needed a tranquilizer myself,” she jokes.

“No, I didn’t,” she quickly adds. “But it isn’t what I would have done two weeks ago. And the patient is well aware she’ll have chemotherapy and radiation, and she’s well aware this may or may not be the new standard.”

“We need time to digest this,” says Dr. Rosenthal. But in their practices, clinicians are also having to respond immediately, not so much digesting the information as consuming it quickly, like a couple of 7-Eleven taquitos.

So, in the meantime, she keeps adding items to her things-to-worry-about list. “You do wonder if, 15 years from now, we’re going to see a much higher percentage of axillary failures. Right now we’re not seeing an excess of axillary recurrences when we do lesser operations. But we may just need to follow patients longer term. There may be a tradeoff that we don’t see yet.”

“These are the things that keep me up at night,” she continues. “What surprising things am I going to see before I retire?”

Both papers are also launching discussions about the usefulness of intraoperative assessment. “When I mentioned it to one of my favorite pathologists, he said, ‘Great—I hate those frozen sections.’ It puts a lot of pressure on the pathologists,” Dr. Rosenthal says.

This would be an easy sell to his pathology colleagues, Dr. Weaver agrees. “They would love not to do an intraoperative assessment on sentinel nodes.” The search for micrometastases and ITCs has boosted false-negative rates. “And pathologists feel horrible when they’ve missed something. Although, when I look at the science, I’m probably one of the few people in my group who doesn’t feel horrible about our false-negative rates going up.”

What might be useful, he says, would be to cut the node in half once it’s removed to look for obvious tumor. If there isn’t any, it might be better to defer to permanent section analysis. That would allow pathologists to convey complete information about the tumor and the nodes, rather than making judgments based on positivity/negativity.

Even with the heft of B-32 to support him, Dr. Weaver recognizes no answers are final. “We keep proving we can do less,” says Dr. Weaver. “But we can do less only in the context of appropriate therapy.

“Sometimes people lose sight of that,” he continues. “I think if you had done this exact same study [B-32] 30 years ago, you would not have gotten the same results. Because our treatment would not have been as good.”

Dr. Weaver even finds some humor in the continual flips of the occult metastases debate, which began more than 60 years ago (Saphir O, Amromin GD. Cancer. 1948;1:238–241). “What will happen 10 years from now,” he predicts, “is that someone will look at B-32 and decide it wasn’t good enough. And they’ll design something new.”