Researchers Debate Need for Radiotherapy After Excision In Subset of DCIS Patients

SAN ANTONIO—Is adjuvant radiotherapy necessary for all breast cancerpatients undergoing breast-conserving therapy? Prospective, randomizedstudies from the NSABP indicate that it is, Norman Wolmark, MD, said ina minisymposium held in conjunction with the San Antonio Breast Cancermeeting.

But based on a retrospective study from The Breast Center, Van Nuys,Calif, incorporating the Van Nuys Prognostic Index, Melvin Silverstein,MD, believes there is a DCIS subset that can safely forgo radiotherapy.

Dr. Wolmark, of the the University of Pittsburgh and chairman of theNSABP, said that a randomized prospective NSABP trial, B-21, designed todetermine whether tamoxifen (Nolvadex) could replace radiotherapy in womenwith very small, noninvasive tumors, has had "abysmal" accrual,because US physicians don't want to give up radiotherapy.

Until such trials are completed, Dr. Wolmark recommends basing treatmenton data from the B-17 protocol, which randomized 818 patients to lumpectomyalone (with free margins) or lumpectomy plus radiotherapy to the ipsilateralbreast. To date, these patients have been followed for an average of 7.5years.

The most recent analysis continues to show that "radiotherapy significantlyand unequivocally reduces the incidence of subsequent ipsilateral noninvasiveand invasive cancer," Dr. Wolmark said.

He noted that in this study, tumor size did not predict for recurrence.Margin status and comedo necrosis were independent prognostic discriminants,but in patients with uninvolved margins who received radiotherapy, comedonecrosis did not increase the risk of ipsilateral tumor recurrence. Inother words, he said, the addition of radiotherapy wipes out comedo necrosisas a prognostic discriminant when margins are not involved.

Randomized Trials

"It is fine to hypothesize about possible biologic interactionsthat could identify a subset of patients who do not benefit from radiotherapy,"he said, "but we must not be deterred from using randomized prospectivetrials to answer the very basic questions relative to DCIS."

Dr. Silverstein, director of The Breast Center at Van Nuys, maintainedin his talk that "subset analysis is important." He noted thatB-17 was designed to show only whether radiotherapy "works" inpreventing recurrence in DCIS. "It does work," he said, "butnow we want to know which patients it works in, how well it works, andif there are patients who don't need radiation."

In his series, 418 DCIS patients treated with excision alone (219) orexcision plus radiotherapy (199) have been followed for almost six years,with data projected to seven years. "These patients are highly selectedand nonrandomized; they chose the treatment they wanted," he said.

There have been 66 local recurrences in the breast to date, 95% at ornear the primary cancer. "This hammers home the point that, in alllikelihood, these patients were inadequately excised," he said.Of these recurrences, 45% were invasive, with five distant metastases,"much more than I would like."

Subset Analysis

To do a subset analysis, the researchers reviewed 30 prognostic factors;by multivariate analysis, only three were found to be significant: tumorsize, margins, and the Van Nuys classification, made up of two biologicmarkers, comedo necrosis and nuclear grade.

In this system, lesions are classified as group 1 (low or intermediategrade, no comedo necrosis), group 2 (low or intermediate grade, with comedonecrosis), and group 3 (high grade with or without comedo necrosis).

The cases were then divided by size and margins into groupings that"made statistical sense and gave three different outcomes," Dr.Silverstein said. Tumors 15 mm or less were designated small; 16 to 40mm, intermediate; and 41 or more, large. Margins of 10 mm were consideredwide; 1 to 9 mm, intermediate; and 1 mm, narrow.

They then devised a simple for-mula, assigning a score for histopathologicclassification (1 point for group 1, 2 points for group 2, and 3 pointsfor group 3); tumor size (1 point for small, 2 for intermediate, and 3for large); and margins (1 point for wide, 2 for intermediate, and 3 fornarrow). "So the best a tumor can be scored is 3 and the worst is9," he said.

The Van Nuys series, scored by this system, broke down into three statisticallysignificant groups: Those with a score of 3 to 4 had a very low recurrencerate; those with a score of 5 to 7 had an intermediate recurrence rate;and those scoring 8 to 9 had a high recurrence rate.

The most important point, Dr. Silverstein said, is that among the 119patients who scored between 3 and 4, outcome did not differ significantlywhether or not they received radiotherapy. Among the 259 patients who scored5 to 7, radiotherapy decreased the local recurrence rate by 13%.

In the high-risk group scoring 8 to 9 on the index, five-year localrecurrence rates were very poor: 60% with radiation and 100% without.

Dr. Silverstein noted that the data need to be prospectively verified,but he believes a simple guideline can be derived from these results, namely:For a score of 3 to 4, consider excision only; for a score of 5 to 7, considerexcision plus radiotherapy; for a score of 8 to 9, consider mastectomy.

Thorough Tissue Processing

Dr. Silverstein emphasized that to use the Van Nuys Prognostic Index,researchers must process the tumor tissue in the same thorough manner asis done at The Van Nuys Breast Center. In their protocol, stereotacticcore biopsy is performed first to determine malignancy, and he calls thisthe key to complete excision. "If you know the diagnosis, then you'renot afraid to be aggressive in the amount of tissue you remove," hesaid.

Typically, four wires are used to bracket the entire lesion (see below), which is then removed with a wide excision. Hemoclips areplaced on the specimen, and multiple view specimen radiograms are takento determine that the margins are clear. The specimen is then color codedon all six surfaces and sectioned at 2 to 3 mm intervals. Such thoroughsectioning is necessary to determine the true size of the tumor and torule out the presence of microinvasion.

In answer to a question from the audience, Dr. Silverstein acknowledgedthat this type of pathology processing is expensive, running about $300to $400 per case, and is becoming financially more difficult to do in themanaged care era.

Dr. Wolmark added that implementing the Van Nuys protocol in today'senvironment "would be an enormous task," but, he said, if thevalue of the Van Nuys system were confirmed in a randomized, prospectivetrial, "I think we would find a way to implement all the techniquesnecessary to support that approach."

At present, however, in Dr. Wolmark's opinion, this approach remainsa hypothesis that "cannot be recommended as a universal algorithmfor the treatment of DCIS, particularly since it is not concordant withthe findings of the NSABP."