Bergström R, Adami H O, Gustafsson L, Pontén J, Sparén P
Department of Statistics, Uppsala University, Sweden.
J Natl Cancer Inst. 1993 Jul 7;85(13):1050-7. doi: 10.1093/jnci/85.13.1050.
Cytologic screening and follow-up can reduce the incidence of cervical cancer by detection and removal of precursor lesions. It is unknown, however, whether differences in histopathologic criteria for these precursor lesions affect the benefit of screening. These criteria may be difficult to study, but they are likely to be reflected in reported incidence of in situ cancer in small areas of Sweden.
Our purpose was to test the hypothesis that the benefit of screening can be predicted by histopathologic criteria as reflected in the reported incidence of cancer in situ.
Incidence data were from the Swedish National Cancer Registry. Regression models showing the relationship between in situ and invasive cancer were formulated and estimated. Each county (total, 24) was a unit of measurement, and adjustment was made for the incidence of invasive cancer before screening.
During population-based screening in Sweden, the incidence of cancer in situ varied about fourfold among the 24 counties, which indicates that the criteria used to diagnose cancer in situ differed markedly. No statistically significant (P < .05) associations were found between the incidence of cancer in situ in 1965, 1970, or 1975 and the reduction in invasive cancer 5, 10, or 15 years later. According to the best-fitting model, detection of 100 extra cases of cancer in situ per 100,000 women per year in 1975 resulted in a reduction of 1.0 (95% confidence interval [CI] = -1.6-3.7) cases of invasive cancer 10 years later. The corresponding best model estimate implied a reduction of 4.6 cases (95% CI = 1.5-7.7) in a model restricted to cancer in situ in patients aged 20-50 years (when organized screening took place), invasive cancer in patients aged 30-60 years, and cancer in situ measured in 1970.
The absent, or at most weak, association between detection of cancer in situ and subsequent reduction in invasive cancer indicates that relaxed histopathologic criteria for cancer in situ may result in extensive, unnecessary treatment of lesions that would regress spontaneously.
Further studies are urgently needed to enable identification of neoplasms likely to progress to invasive, fatal disease.
细胞学筛查及后续跟进可通过检测和切除癌前病变来降低宫颈癌的发病率。然而，这些癌前病变的组织病理学标准差异是否会影响筛查的益处尚不清楚。这些标准可能难以研究，但它们可能会反映在瑞典小区域原位癌的报告发病率中。
我们的目的是检验这一假设，即筛查的益处可通过原位癌报告发病率所反映的组织病理学标准来预测。
发病率数据来自瑞典国家癌症登记处。构建并估计了显示原位癌与浸润癌之间关系的回归模型。每个县（共24个）为一个测量单位，并对筛查前浸润癌的发病率进行了调整。
在瑞典基于人群的筛查期间，24个县中原位癌的发病率相差约四倍，这表明用于诊断原位癌的标准存在显著差异。在1965年、1970年或1975年的原位癌发病率与5年、10年或15年后浸润癌发病率的降低之间未发现具有统计学意义（P <.05）的关联。根据最佳拟合模型，1975年每10万名女性中每年多检测出100例原位癌病例，10年后浸润癌病例减少1.0例（95%置信区间[CI] = -1.6 - 3.7）。在一个仅限于20 - 50岁患者（进行有组织筛查时）的原位癌、30 - 60岁患者的浸润癌以及1970年测量的原位癌的模型中，相应的最佳模型估计显示减少4.6例（95% CI = 1.5 - 7.7）。
原位癌的检测与随后浸润癌发病率降低之间不存在关联，或至多存在微弱关联，这表明对原位癌放宽组织病理学标准可能导致对那些会自发消退的病变进行广泛且不必要的治疗。
迫切需要进一步研究，以确定可能进展为浸润性致命疾病的肿瘤。
