Experts peg MD Anderson breast radiotherapy study as flawed, misleading

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
Thumbs down, rejection - 120.95 Kb

Four experts attempted to set straight the record on accelerated partial breast brachytherapy (APBI) after a Dec. 7 presentation by MD Anderson researchers at the San Antonio Breast Cancer Symposium, which concluded that APBI is associated with inferior effectiveness and increased toxicity compared with whole breast irradiation (WBI) in older women. The researchers outlined numerous flaws ranging from the data source to study design to omitted information, during a teleconference on Dec. 13.

APBI treats only the part of the breast affected by cancer and the treatment time is decreased from several weeks to four or five days.

APBI is associated with double the rate of mastectomy compared with WBI at 4 percent and 2.2 percent, respectively, reported Grace L. Smith, MD, of the University of Texas MD Anderson Cancer Center in Houston, and colleagues at the San Antonio meeting. The MD Anderson group also linked APBI with higher risks of hospitalization and infection as well as higher five-year incidences of rib fracture, fat necrosis and breast pain.

The findings shared at the San Antonio Breast Cancer Symposium, however, don’t tell the entire story, contended a group of experts.

“This study of an inherently inaccurate database with no data on tumor characteristics and margin status and questionable outcomes and nonvalidated surrogate endpoints should be looked at with appropriate skepticism in the face of 20 years of retrospective studies and two large prospective randomized trials to the contrary,” said Peter D. Beitsch, MD, co-principal investigator of the American Society of Breast Surgeons' MammoSite Registry.

Smith and colleagues based their claims on an analysis of 130,535 women in the Surveillance Epidemiology and End Results (SEER) database treated with conservative surgery followed by APBI versus WBI between 2000 and 2007.

One of Smith and colleague’s design flaws is the use of SEER, said Beitsch. SEER, a Medicare claims database, is typically used for economic research. The database contains no specific clinical data on prognostic factors such as tumor margins or characteristics or marker status, such as estrogen or progesterone receptors and HER2.

“APBI is one the most studied treatments in history,” said Robert Kuske, MD, co-principal investigator, NSABP B-39 study. Nine randomized controlled trials have compared APBI to WBI.

In contrast to the tight control of randomized clinical trials, the current report, which has not yet been published, is a retrospective look at Medicare patients who underwent APBI at the discretion of thousands of physicians. Among this group of patients, physicians based treatment decisions on cancer extent, grade, surgical margins and other factors such as obesity, diabetes and age. “This study failed to take these important tumor and patient issues into account, and is therefore biased,” said Kuske.

Smith et al also used subsequent mastectomy as a validated surrogate for local failure. No published literature has validated this as a surrogate, according to Beitsch.

Beitsch questioned Smith and colleagues’ calculations, calling the two-fold increase in mastectomy “quite misleading.” He noted 4 percent is not the double of 2.2 percent, adding, “Either way, [the mastectomy rates for both procedures] are quite small and certainly acceptable after lumpectomy and radiation.”

Kuske noted the discrepancy between statistical significance and clinical importance, explaining that it may be easy to achieve statistical significance in a study of this size. However, the difference is of questionable clinical importance to individual patients.

Smith and colleagues did not report on recurrence rates while admitting that survival outcome among the two treatments were equivalent. The recurrence rate for lumpectomy without radiation therapy ranges from 35 to 40 percent, said Kuske.

Final flaws in the MD Anderson report related to the researchers’ measures for complications. According to Beitsch, there is no standard definition for fat necrosis, and the researchers did not define breast pain. He questioned how breast pain might be measured in the SEER database.

“The other glaring data point is the 9.6 percent hospitalization rate,” Beitsch said. The researchers did not link hospitalization to a diagnosis or provide a time period. “It’s worth noting that APBI is often used in older, sicker patients who may not be candidates for six to seven weeks of whole breast irradiation,” continued