| Summary: | 碩士 === 國立臺灣大學 === 預防醫學研究所 === 94 === Purpose: To evaluate effectiveness of hospital-based breast cancer surveillance with periodic mammography/ultrasonography examinations; previous breast cancer screening study was conducted on population-based randomized trials, which was distinct from breast cancer surveillance at hospitals.
Materials and methods: We used cancer registry file from one cancer prevention center between 1998 and 2004 to ascertain 841 breast cancer cases. Surveillance history was obtained through administration data. Healthy controls were those undergoing breast image examinations and were free of cancer at the same institute concurrently. The whole breast surveillance cohort was analyzed using logistic regression model stratified by age group (>50 versus <50 years) with dichotomous responses. Subgroup analysis of cancer patients were targeted for clinical and pathological features during diagnosis of breast cancer; including tumor size (>2cm versus <2cm), axillary lymph node status (presence versus absent), and histological behavior (invasive cancer versus carcinoma in situ). These surrogate endpoints were validated from disease-free survival and mortality during follow-up (log-rank test: p<0.001 and p=0.0343, respectively). Discrete states time homogenous Markov model was applied to histological behavior; effects of surveillance as transitional parameters reduction were surveyed.
Results: Whole cohort analysis showed that for pre-menopause women (age<50 years), ultrasonograophy alone, rather than mammography or two modalities combined, recognized more benign breast conditions than cancer. The effect was less prominent for older women (age>50 years) and was only observed during the preceding one to two years (odds ratio of breast cancer: 0.179, p=0.0041). Subgroup analysis of breast cancer patients showed that mammography alone between one to two years before cancer diagnosis had protective effect upon histological behavior (age-adjusted odds ratio of invasive cancer: 0.048, p=0.016). Three states Markov model showed net efficacy of surveillance was 66.18% (63.44~72.10%); surveillance within 2 years of disease onset resulted in 65.16% (59.79~76.74%) reduction of carcinoma in situ to invasive cancer transition.
Conclusion: Current study highlighted hybrid nature of hospital-based breast cancer surveillance. Compared with mammography, ultrasonography detected more benign breast conditions. Mammography alone remains as gold standard for breast cancer screening. Further study to elucidate the nature and performance of hospital-based breast cancer surveillance is warranted.
|
|---|
