Active Surveillance for DCIS Moves Closer to Standard Option but ‘Not Quite There’

Women who opted for active surveillance (AS) for ductal carcinoma in situ (DCIS) had ipsilateral invasive breast cancer survival (iiBCS) similar to patients who had immediate surgery, a randomized trial showed.

After a total follow-up of about 4 years, iiBCS favored AS at 24 months (94.3% vs 90.6%) and met statistical criteria for non-inferiority at 36 and 48 months. Invasive cancers detected in women on AS were larger but no more aggressive than the invasive cancers in the surgery group.

An earlier analysis after 22 months of follow-up favored immediate surgery. The disparate results at different time points suggest a different role in DCIS for surgery and AS, reported Jelle Wesseling, MD, of the Netherlands Cancer Institute in Amsterdam, at the European Breast Cancer Conference (EBCC) in Barcelona.

“Probably this is a matter of delayed detection in active surveillance, so that primary surgery is, in fact, a diagnostic net, catching away early cancers that are not diagnosed yet on mammography because they might be small,” said Wesseling. “Does this have consequences for how intense your treatment is, or outcome? I guess you know the answer, but we’re working on this to find out. We don’t have the science yet, then this will make a difference. Maybe I can share that with you in 2 years or so.”

Another study reported at EBCC failed to provide clear support for AS in DCIS. The much smaller LORIS trial showed a 5-year ipsilateral iBCS survival of 94.2% with immediate surgery versus 87.5% with AS. The upper limits of the confidence intervals fell outside the original 2.5% margin for non-inferiority but within the 5% margin subsequently found to be acceptable, said Matthew Wallis, MBChB, of the University of Cambridge Hospital in England.

“We have insufficient data to conclude that active monitoring is non-inferior to surgery for low-risk DCIS, but our hazard ratio point data does fall within the 5% acceptable non-inferiority margin,” said Wallis. “I suppose if we had a larger sample size, we might have got to that point. We do know that LORIS [trial] central histological review was able to identify a low-risk population, with no detriment to psychological well-being. We are providing that information to patients who might want to avoid surgery.”

Need for Better Patient Selection

The two current trials, plus two other recent studies of AS for DCIS, have yielded inconsistent results, leaving the issue unresolved, said invited discussant Shelley Potter, MBChB, PhD, of the University of Bristol in England. DCIS is a heterogeneous disease, and current technology cannot identify the low-risk subgroup most appropriate for AS.

“I think we all agree that low-risk DCIS doesn’t need treatment and that active monitoring is safe,” said Potter. “We need to work out who those patients are…We’re getting there. We’re just not there yet.”

Interest in AS for DCIS has come from data showing that an estimated 70-80% of untreated DCIS will not progress to iBC, an observation that Wesseling said has annoyed him for 15 years or more.

“There’s quite some overtreatment at stake, because DCIS, in and of itself, is not lethal, and the survival with treatment is really good — maybe related to overtreatment, as well,” he said. “Many women carry the burden of probably needless treatment.”

To date, four clinical trials have evaluated AS for DCIS, including the two reported at EBCC. The randomized COMET trial, involving about 1,000 patients with low-risk DCIS, showed no significant difference in the rate of iBC between AS and guideline-concordant care after 2 years of follow-up but a trend in favor of AS. The single-arm LORETTA trial from Japan enrolled 337 patients into AS and showed a 5-year incidence of iiBC of 9,8%, exceeding the 7% margin for non-inferiority.

Latest Data

LORD began as a randomized trial, allocating women with DCIS 1:1 to immediate surgery or AS, but accrual did not work out as planned, because “the women had strong preferences,” said Wesseling. As a result, enrollment switched to patient-preference accrual. Additionally, investigators limited enrollment to patients with hormone receptor-positive grade 1 or 2 DCIS. Subsequently, the trial accrued 1,914 patients, 1,365 accepted into the trial. The surgery arm had 333 patients and the AS arm had 1,032.

The trial had a prespecified interim futility analysis after 60 iiBC events. After a median follow-up of 22.4 months, the analysis showed an estimated 36-month iiBCS of 99.1% in the surgery arm versus 89.9% in the AS arm, resulting in a hazard ratio of 6.303. However, Wesseling noted that the analysis excluded 33 iBCs found at primary surgery.

The data and safety monitoring committee recommended stopping enrollment but continuing follow-up to 160 iiBC events. After additional follow-up of 24 months, a second analysis was performed, this time including the invasive cancers found at primary surgery. The analysis showed a trend favoring AS at 24 months (iiBCS 94.3% vs 90.6%) and non-inferiority at 36 months (89.9% vs 90.6%) and 48 months (86.6% vs 88.4%).

Median tumor size in the AS group was 9 mm as compared with 6 mm in the surgery arm, but a similar proportion of invasive cancers were grade 1/2 in the AS arm (89.1%) and the surgery arm (92.9%).

Wallis reported findings from the LORIS randomized trial, which involved 381 patients with grade 1/2 DCIS allocated to surgery or AS. The trial initially had a recruitment goal of 932 patients but was revised to 188 because of slow accrual. The primary outcome was iiBCS at 5 years.

Wallis pointed out that 23 patients assigned to surgery switched to AS, as compared with only one patient switching from AS to surgery. As a result, analysis was performed on the “as-treated” population, 69 in the surgery group and 112 in the AS group.

The 6.7% absolute difference in iiBCS translated into an HR of 1.54, which exceeded the HR of 1.45 for a 2.5% non-inferiority margin but was well within the HR of 1.91 associated with a 5.0% margin.