Does breastfeeding affect the risk of gastric cancer?
2003; Wiley; Volume: 106; Issue: 6 Linguagem: Inglês
10.1002/ijc.11291
ISSN1097-0215
Autores Tópico(s)Helicobacter pylori-related gastroenterology studies
ResumoInoue et al. 1 recently studied relations between reproductive and menstrual factors and the risk of gastric cancer, considering case-control data from Japan. According to their working hypothesis, the lower risk of gastric cancer seen in women may in part be linked to factors involving fertility and pregnancies. Indeed the data of Inoue et al.1 indicated that parous women with a short breastfeeding period per child experienced a lower risk than nulliparous women, although women with a long period of breastfeeding had a considerably higher risk. The positive association with the average duration of breastfeeding per child was especially pronounced for cancers in the upper third of the stomach. We have previously published summary estimates of potential associations between duration of breastfeeding and incidence of gastric cancer, considering follow-up data from 3 Norwegian counties.2, 3 To our knowledge, there are no other reports on this relation except the recent one by Inoue et al.1 Prompted by the results in the Japanese study, we have now reanalysed our data in more detail on the basis of an extended follow-up. Our cohort included female residents of the counties of Nord-Trøndelag, Aust-Agder and Vestfold, born in 1886–1928 and living mostly in rural areas, largely reflecting breastfeeding habits different from those seen later. Information on pregnancies and breastfeeding habits was collected at a screening for breast cancer during 1956–1959. The cohort was followed in the period 1961–1998 with regard to cancer diagnoses reported to the Cancer Registry of Norway and with regard to deaths and emigrations. To avoid possible diagnostic problems in very old women, our analysis took into account the follow-up for each person only until the age of 80 years. Relations between the risk of gastric cancer and reproductive factors may be different in women aged less than 50 years3 and in any case the study of Inoue et al.1 included postmenopausal women only. Thus our present analyses were restricted to the age interval 50–79 years. Our cohort included 48,225 parous women with data on breastfeeding contributing person-years to this age interval and similarly 11,415 nulliparous women. The mean duration of breastfeeding per child among parous women was 6.0 months, with a standard deviation of 4.0 months. During follow-up, a total of 605 cases of gastric cancer were diagnosed. The information recorded at the Cancer Registry of Norway allows a classification of gastric cancer into 4 categories with regard to subsite:4 proximal stomach (cardia and fundus; 71 cases), middle stomach (body; 65 cases), distal stomach (antrum and pylorus; 190 cases) and subsite unknown (both curvatures, overlapping and unspecified subsites, and previous gastric resection; 279 cases). Relative incidence was assessed by Poisson regression analysis of log-rates, considering the number of person-years contributed by each woman during follow-up. Table I shows results in the categories of average duration of breastfeeding per child considered by Inoue et al. 1 The reference category corresponds to 7–12 months of breastfeeding rather than ≥13 months, however, as few cases occurred in the latter category. Table I also includes results from linear analyses based on 46,132 parous women with known age at first delivery, with adjustment for parity and age at first delivery. Further adjustment for age at last delivery in separate analyses among multiparous women did not essentially affect the risk estimates. Our results do not provide any firm support to the hypothesis that the risk of gastric cancer increases with the average duration of breastfeeding. Although the estimated linear trends for all subsites combined and for the proximal subsites reflected positive relations, the trends were quite weak. However, nulliparous women showed risk estimates intermediate between those for the different breastfeeding categories. Our data permitted only a crude estimation of risk in women with a very long average period of breastfeeding. It is an open question whether the higher risk seen in this category in the Japanese data of Inoue et al.1 can partly be related to other factors in the population considered. No definite trend was found in our data that compared risk estimates for different subsites, but considering the width of the confidence intervals, our estimates may be consistent with a tendency to a moderate positive relation among cases in the upper part of the stomach only. The Norwegian classification into subsites differs somewhat from that used in the Japanese data of Inoue et al., 1 in particular for the large group that represents an unknown subsite. It is unlikely, however, that imprecise subsite classification is related to the potential risk factors considered. The proximal subsites of gastric cancer probably include cancers related etiologically to adenocarcinoma of the esophagus.4 However, in a case-control study of such cancers, an inverse association was found with breastfeeding.5 Overall, the relations suggested by various studies between risk of gastric cancer and reproductive and menstrual factors present a rather bewildering picture, especially for subgroups. More detailed studies are clearly called for. It is still noteworthy that Inoue et al.1 found a similar inverse U-shaped relation with age at menopause as seen in our data3 and in a previous study.6 An inverse relation with age at first delivery for cancers of the distal stomach was also indicated in both data sets. Yours sincerely, Ivar Heuch, Gunnar Kvåle
Referência(s)