Mieke Kriege
Breast Cancer Screening in Women with a Familial or Genetic Predisposition The role of MRI
15 november 2006
Promotor: Prof. dr. J.G.M. Klijn
Co-promotor: Dr. C.T.M. Brekelmans
Women with a strong family history of breast and/or ovarian cancer combined with young ages at diagnosis of affected family members have an increased risk of these types of cancer. In 1994 and 1995 respectively, the BRCA1 and BRCA2 genes were identified. A germline mutation in one of these genes is associated with very high risks of early onset breast and ovarian cancer.
Current options for BRCA1/2 mutation carriers to reduce their risk of breast cancer or death by breast cancer include prophylactic mastectomy, prophylactic salpingo-oophorectomy, chemoprevention and screening. Screening for breast cancer is also offered to women with a familial predisposition, but without a proven BRCA1/2 mutation. Several studies have investigated the efficacy of mammographic screening, sometimes in combination with clinical breast examination (CBE)) in high-risk groups of women. However, the efficacy of mammography screening has never been clearly demonstrated. Sensitivity of mammography was low this group of women in comparison with post-menopausal women screened in population based studies, most likely because of the young screening age of and consequently frequent a high density of the breast tissue. MRI appeared to be a sensitive method for detection of breast cancer in a diagnostic setting. For this reason, in the late nineties several breast cancer screening studies comparing the value of MRI and mammography were set up in women with a genetic susceptibility. Results of pilot and preliminary studies showed in all of them a very high sensitivity of MRI, while sensitivity of mammography was never higher than 50%. Recently, the first results of four large prospective studies were published, among which the Dutch national MRISC study. In this thesis, the short-term results of the MRISC study are described. Two of the main objectives of the MRISC study are addressed in this thesis:
- Assessment of the efficacy of screening in diagnosing early-stage breast cancer in women with a familial or genetic predisposition
- Assessment of the value of MRI in this screening scheme compared to mammography
In Chapter 3 the design of the MRISC study is described. The start date of this study was 1 November, 1999, and 6 cancer centers and university hospitals participated in this study. In these centers, women who were already under intensive surveillance and women who came for screening for the first time were asked by a clinician to participate in the study. The inclusion criteria for participation were a cumulative lifetime risk of breast cancer of >15% due to a familial or genetic predisposition, an age between 25 and 70 years, or younger than 25 in women from families with a very young age of onset (<30 years). Women with previous breast cancer or symptoms of breast cancer were excluded. The participants visited the hospital twice a year for a clinical breast examination and once a year for a mammography and MRI, with a maximum period of 6 weeks between both imaging modalities. Mammography and MRI were independently evaluated. Breast cancer detection rates, screening parameters of mammography and MRI, and characteristics of tumors diagnosed in the MRISC study are described in Chapter 4. In total 1909 eligible women were entered in the MRISC study, including 358 carriers of a germline mutation. During a median follow-up time of 2.9 years, 51 tumors were detected, of which 1 lymphoma, 44 invasive breast cancers and 6 ductal carcinomas in situ (DCIS). The detection rate for breast cancer was 9.5 per 1000 women-years at risk, with the highest detection rate for mutation carriers (26.5 cancers per 1000 women-years at risk). MRI was a more sensitive screening method than mammography (71.8% vs. 33.3%, respectively); the sensitivity of clinical breast examination was only 17.9%. However, mammography was more sensitive than MRI for detecting DCIS: mammography detected 5 out of 6 cases of DCIS, MRI only one out of 6. Overall, the specificity was higher for mammography (95.0%) than for MRI (89.8%); the specificity for clinical breast examination was 98.1%. Sensitivity and specificity were combined in a receiver operating characteristic (ROC) curve. The area under the curve for MRI was significantly higher than for mammography, meaning that MRI can better discriminate between women with and without breast cancer. Characteristics of tumors detected in the screened group were compared with those of two age-matched symptomatic control groups. Tumors detected in the MRISC study were detected in a more favorable stage than those of the symptomatic controls, they were more often .1 cm (43.2 vs. 14.0% (P<0.001) and 12.5% (P=0.04) in both control groups, respectively, and the incidence of positive lymph nodes was lower in the MRISC study group (21.4%) compared to 52.4% (P<0.001) and 56.4% (P=0.001) in the two control groups.
Screening parameters and tumor characteristics can differ between the first round and subsequent rounds of a screening program. In the MRISC study, yearly MRI is added to the existing screening program of yearly mammography and 6-monthly clinical breast examination. This means that the results described in chapter 4 are mainly based on comparing a first MRI with a mammography after a prior mammography. As a result the differences in sensitivity and specificity between MRI and mammography might be overestimated. In chapter 5 we have described whether the differences in sensitivity and false positive rate between MRI and mammography as observed in chapter 4 are maintained in the subsequent rounds. Although we observed the highest difference in both sensitivity and false positive rate between MRI and mammography in the first round, also in the subsequent rounds these statistically significant differences between MRI and mammography remained with respect to screening parameters: 76.5% (MRI) vs. 29.4% (mammography) P=0.02 for sensitivity, 8.2% (MRI) vs. 4.6% (mammography); P<0.001 for the false positive rate. No major differences with respect to tumor stage were found between tumors detected at the subsequent rounds in comparison with those detected in the first round. In the subsequent rounds tumor characteristics remained more favorable with respect to size and lymph node involvement than in symptomatic age-matched controls. Because results of these subsequent rounds were possibly most predictive for long-term effects, it is expected that on the longterm this screening program will ultimately contribute to a decrease of breast cancer mortality.
In chapter 6 we investigated more in detail the contribution of MRI in the early detection of breast cancer reported in chapter 4. We did this by investigating the number and percentage of tumors that were detected by MRI and missed by mammography and by comparing tumor characteristics of MRI-only detected tumors with those of other screen-detected tumors. Forty-five of the 50 detected breast cancers were evaluable for detection method and were included in the analysis of chapter 6. Twenty-two (49%) of the 45 breast cancers were detected by MRI and were not visible at mammography, of which 20 (44%) were also not palpable (MRI-only detected tumors). Eight breast cancers (18%) were detected by mammography and not visible on MRI, 10 (22%) breast cancers were detected by both MRI and mammography. One breast cancer was only detected by clinical breast examination and missed by mammography and MRI, and 4 (9%) were detected in the interval between two screening visits. The MRI-only detected tumors were more often .1 cm than all other screendetected cancers (58 vs. 31%; P=0.11). Also MRI-only detected tumors were more often node-negative than other screen detected cancers (94 vs. 59%; P=0.02). This suggests that MRI importantly contributes to the early detection of breast cancer, as described in chapter 4.
Whether age, a BRCA1/2 mutation, menopausal stage and breast density independently influenced sensitivity and false positive rate (1-specificity) of mammography and MRI is investigated in Chapter 7. An unexpected finding was that sensitivity of MRI was significantly decreased in women with a high mammographic breast density compared to women with a low breast density (adjusted OR (OR adj) 0.08 [95% CI 0.01-0.84]). As expected, also the sensitivity of mammography was decreased by a high breast density, but not significantly (OR adj 0.42 [95% CI 0.10-1.7]). Further there were non-significant trends of a decreased sensitivity of mammography in younger and pre-menopausal women and in women with a BRCA1/2 mutation. These trends were not observed for MRI. False-positive rates of both mammography (OR adj 1.67 [1.22-2.28]) and MRI (OR adj 1.21 [0.97-1.51]) were increased by high breast density, that of MRI by pre-menopausal status (OR adj 1.70 [1.23- 2.36]), young age (OR adj 1.58 [1.17-2.13]) for women 40-49 years versus women ≥50 years).
In chapter 8, the tumor growth rate of 30 tumors detected in BRCA1/2 mutation carriers is compared with the growth rate of 25 tumors detected in women with a familial or genetic risk, but no proven BRCA1/2 mutation. An exponential tumor growth was expected and growth rate was expressed in tumor volume doubling time. Tumor size was measured on either MRI or mammography. The tumor volume doubling time was 45 days (95% CI 26-73 days) in BRCA1/2 mutation carriers and this was significantly shorter (P=0.048) than in women without a proven mutation, in which the tumor volume doubling time was 84 days (95% CI 58-131 days) (ratio 0.5). However, when this result was adjusted for age and menopausal status, there was only a non-significant trend for a faster tumor volume doubling time in mutation carriers compared to non-mutation carriers (adjusted ratio 0.7 [95% CI 0.4-1.3]). Only a higher age significantly prolonged tumor volume doubling time (adjusted ratio 1.9 per 10 years older age [95% CI 1.1-3.4]).
In chapter 9 results described in this thesis are discussed and conclusions are summarized. Further, recommendations for screening of this group and for further research were done. The overall conclusion was that MRI is a more sensitive screening method than mammography, but less specific. MRI is a more accurate method and can better discriminate between women with and without breast cancer. Screening by the scheme used in the MRISC study (yearly mammography and MRI and 6-monthly breast examination) facilitated an early breast cancer diagnosis. Therefore, in the current screening guidelines for BRCA1/2 mutation carriers, yearly MRI is added to the previously used screening scheme of yearly mammography and 6- monthly clinical breast examination. In women with an increased risk but without a proven BRCA1/2 mutation, the breast cancer incidence is lower and the sensitivity of mammography is better than in gene mutation carriers, causing a less favorable cost-effectiveness for the addition of MRI. Therefore, for this group of women, currently there is no consensus whether MRI has to be advised as standard screening tool and MRI screening should preferably be performed in a research setting. An important research question for the future is whether highrisk women without a proven BRCA1/2 germline mutation should be screened by MRI, and what should be their minimum genetic risk and age at entry.