Prediagnostic Hematocrit Values and Subsequent Cancer Risk1

In this prospective study, the association between hematocrit values measured in 1965-68 and subsequent cancer incidence was studied among 7737 Japanese-American men in Hawaii. With an increase in hematocrit levels, there was an increasing risk for lung cancer, especially squamous or small cell type, and for kidney cancer. However, a statistically significant trend remained only for kidney cancer after adjusting for smoking history. The relative risk for kidney cancer was 4.94 (95% confidence interval, 1.03-23.63) for subjeds with hematocrit values of 47 or higher compared with those with hematocrit values of 43 or lower. In contrast, the risk of oral-pharyngeal cancer decreased with increasing hematocrit levels. With adjustment for cigarette smoking and alcohol intake, the relative risk for oral-pharyngeal cancer was 0.20 (95% confidence interval, 0.04-0.88) for subjects with hematocrit values of 47 or higher compared with those with hematocrit values of 43 or lower. The association with kidney or oral-pharyngeal cancer was not affeded by the time interval between examination and diagnosis of these cancers ( 10 and >10 years). Erythropoietin produdion by kidney tumors and micronutrient deficiencies in oral-pharyngeal cancer cases may partially account for these results. Introduction Hematocrit levels are affected by various endogenous and exogenous factors, including sex hormones, psychological stress, cigarette smoking, nutritional intake, and exposure to toxic substances (1). These factors have also been implicated in carcinogenesis. Therefore, hematocrit may be a useful biological marker to identify etiological factors of cancer and to evaluate the effects of various exposures. In a ten-year mortality study, Waters et al. (2) reported that subjects with hemoglobin and hematocrit values near the mean had the lowest mortality rate. Subsequent studies found that an excess of cancer deaths occurred among those with low hemoglobin/hematocrit levels and an excess of cardiovascular deaths occurred Received 5/27/91. 1 This work was supported in part by Grant RO1 CA33644 from the National Cancer Institute, National Institutes of Health, Bethesda, MD. 2 Present address: Division of Epidemiology, Aichi Cancer Center Research Institute, 1-1 Kanokoden, Chikusa-ku, Nagoya 464, Japan. among those with high hemoglobin/hematocrit levels (35). Because these studies usually had a relatively short period of follow-up (2-5), the anemia could be a consequence of the cancer. Several specific cancers have been linked to anemia or low hematocrit values. Patients with Plummer-Vinson syndrome, who have iron deficiency anemia, have an increased risk for upper digestive tract cancer (6-9), and patients with pernicious anemia have an increased risk for gastric cancer (10-13). Otherwise, little is known about the association of hematocrit levels with the occurrence of other cancers. An earlier report from the Honolulu Heart Program showed there was no association between hematocrit levels and total cancer deaths after 10 years of follow-up (14). However, the association of hematocrit values with cancer incidence by site has not been reported. These subjects have now been followed for more than 20 years, and the identification of a substantial number of incident cases of cancer permits us to assess the relation of hematocrit levels to cancer incidence by site. Materials and Methods The subjects for this study were American men of Japanese ancestry, born between 1900 and 1919, and residing on the Hawaiian island of Oahu. They were identified by the Honolulu Heart Program in 1965 with the use of the comprehensive 1942 Selective Service draft registration files (15). Of 11,148 identified men, 8,006 (71.8%) were interviewed and examined between 1965 and 1968. One hundred eighty men (1.6%) died before they could be examined, and 2962 (26.6%) did not participate in the program. The subjects came to the baseline examination in a nonfasting state, and blood was drawn 1 h after a 50-g glucose load. Hematocrit levels were determined by the method of Guest and Siler (16). During the interview, information was collected on cigarette smoking history and alcohol intake (ounces/month). Subsequent cases of cancer in the cohort have been identified by continuous surveillance of all general hospitals on Oahu and by linkage with the Hawaii Tumor Registry, a member of the Surveillance, Epidemiology, and End Results Program ofthe National Cancer Institute. Each diagnosed case was confirmed by histological cxamination. Histological types of lung cancer (squamous and small cell carcinoma, adenocarcinoma, and others) were determined based on the classification of the World Health Organization (1 7). The slides from each case were reviewed by one of us. There were 81 prevalent cancer cases diagnosed before examination, 142 suspected cancer incident cases without histological confirmation, and 46 examinees without a hematocrit value. These subjects were excluded from the analysis. As a result, 7737 subjects remained in the study. on June 22, 2017. © 1991 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from 52 Hematocrit and Cancer Table 1 Age.adjusted nl(’an lvv ‘Is ot heniat( )( nit by ( 8 (U er silt’ Site’ of cancer No, Mean I’ Upper digestive.rt’spiratory tra( I Oral-pharynx Esophagus Larynx 79 23 27 25 44. 5 4 3.3 45.1 45.2 0.65 t).t) 3 0.42 0.40 Stomach 203 44.5 0.43 Colon Proximal ((don [)istal e olon 250 80 169 44.8 44.6 44#{149}() 0.55 1)79 0. 38 Rectum 99 45.1) 0.31 Hepato.hiliary-pan( r(’ati( syst(’nh Liver Biliary tract Pancreas 89 29 24 56 44.9 44.2 45.6 44.9 0.52 0.42 t). 1 3 (1.61 Lung Squamous or small cell Other cell types 216 109 71 45.2 45.6 44.6 0.02 <0.01 0.76 Prostate 267 44.9 1). 2 1 Other urogenital organs Uroepithelial Renal cell 10.3 84 18 44.8 44.5 46.4 0.68 0.51 1)02 I ymph-hematopoietic tissue Lymphoma Leukemia 75 37 32 44.7 44.4 45.0 0.89 0.51 0.51 Other miscellaneous ( ant t’rs 87 44.7 11.95 All cancers 1468 44.8 0.1)9