Screening of Women at High Familial Risk of Breast Cancer

Abstract

Evaluation of: Kuhl C, Weigel S, Schrading S et al.: Prospective multicenter cohort study to refine management recommendations for women at elevated familial risk of breast cancer: the EVA trial. J. Clin. Oncol. 28, 1450–1457 (2010). In the EVA trial, Kuhl et al. screened 687 women at high risk of breast cancer for up to 3 years with annual MRI. A total of 27 breast cancers were diagnosed, including 11 DCIS and two local recurrences. There were no interval cancers. This data suggests that MRI screening can be used to downstage breast cancers, but larger studies with longer follow-up periods are required to confirm these findings.

In the past decade, research groups in North America and Europe have reported that the sensitivity of screening MRI for women at high risk of breast cancer exceeds that of mammography or ultrasound [1–3]. It is now generally accepted that MRI has the highest sensitivity of any breast screening modality, but several questions remain. First, who to screen? At present, this is based on age, BRCA mutation status and family history, but there is no consensus on how these factors should be integrated. Other risk factors, such as benign breast disease and mammographic density, do not yet feature in the risk assessment.

No study to date has demonstrated that regular MRI screening reduces mortality below that of conventional mammography and no randomized trial has been conducted. Instead, prospective observational trials have been carried out and researchers rely on surrogate end points. In the study by Kuhl et al., the authors considered small, or noninvasive breast cancers to be a good surrogate for cancer survival. It is expected that the great majority of women with ductal in situ carcinoma DCIS or a small nodenegative cancer (<1 cm) should survive their disease. This is a reasonable assumption, but tumor size is an imperfect surrogate for survival for two reasons. First, the strong relationship between tumor size and survival appears to be attenuated for women with a BRCA1 mutation. Second, is the possibility of overdiagnosis [4]. It is not clear how many cases of DCIS would progress to invasive cancer if they had not otherwise been intercepted with screening MRI. In addition, the MRI cohort included women with nonpalpable cases of DCIS. Recently, Kerlikowske et al. reported that palpability was among the strongest predictive factors for identifying cases of DCIS that are likely to progress to invasive cancer [5]. One is more confident about the expected clinical impact of a small invasive cancer, but even here there is some uncertainty. Close examination of the results of several mammography screening trials suggests that overdiagnosis may be a problem [6,7]. If so, then identifying small cancers may not necessarily result in fewer advanced cancers.

Furthermore, it is not clear what is the ideal screening interval is – most studies to date have been based on a 1-year interval – but this interval has been chosen largely as a matter of convenience. In addition, it is reasonable to question if there is any benefit to adding mammography or ultrasound to a regimen that already includes annual MRI.

Results

In the EVA trial, 687 women who were at elevated risk of cancer were screened with a combination of MRI, ultrasound, mammography and physical examination over a period of approximately 3 years. On average, the women underwent 2.4 rounds of screening. In total, 27 breast cancers were diagnosed (16 invasive and 11 DCIS); of these, 25 were identified by MRI. Mammography added two more cases and there was no marginal utility from adding ultrasound or physical examination. Of the 27 cancers, 21 were of the low-risk type (i.e., either DCIS or node-negative and less than 1 cm in size). There were no interval cancers in the cohort.

Significance

The EVA investigators attempt to evaluate an important clinical issue. I welcome new data that supports the position that we are making progress in screening high-risk women for breast cancer. Unfortunately, the shortcomings of this paper were numerous and sufficiently serious that little can be adduced. The mean follow-up was short – on average, the women were followed for 2.5 years and underwent 2.4 rounds of screening. Approximately 100 women only had one screen, and ten cancers were identified on the prevalent screen. It is not a failure of the screening test if a woman presents at her first screen with node-positive cancer. Of the incident cancers, 11 were in situ and 16 were invasive. Of the invasive cancers, some (we are not told how many) were second primary ipsilateral breast cancers and two were local recurrences. These should have been excluded. This muddling of case groups and end points makes it very difficult to see either the forest or the trees. Of the cases that were ‘missed’ by the MRI, one was a 3 mm microinvasive lesion and one was a case of DCIS.

I am not sure if missing these is as ominous as Kuhl suggests. This is difficult for anyone to judge in the absence or a control group who did not have MRI screening. Approximately half of the women opted for an additional 6-monthly screen. Given that the screening protocol was not assigned at random, it is hard to interpret this additional data.

Future perspective

There are critical weaknesses in this paper, and I suspect that if they had been noted by either the reviewers or the editors, would have resulted in mandatory revisions. We would then be in a position to properly review the evidence. However, the lack of interval cancers in the cohort is an important and relevant end point. Unfortunately, we will have to wait for other cohorts to be evaluated in order to come to conclusions regarding other end points.

Executive summary

• Women with a BRCA1 or BRCA2 mutation or a family history of breast cancer are at increased risk of developing breast cancer.

• Screening MRI has a sensitivity superior to that of mammography in detecting ductal in situ carcinoma or invasive breast cancers.

• In a prospective cohort of 687 high-risk women, 27 cancers were detected. Of these cancers, only three were node-positive.

• No interval cancers were detected in the cohort of women undergoing annual MRI screening.