The Risk for Second Primaries in Gastric Cancer Patients : Adjuvant Therapy and Habitual Smoking and Drinking

Study purpose: To examine whether adjuvant therapy for gastric cancer increases the risk for second primaries, and whether smoking and drinking increase the risk. Subjects and methods: 1,631 patients who were newly diagnosed with gastric cancer and underwent curative operation at Osaka Medical Center for Cancer and Cardiovascular Diseases during 1978-92. Incidence of second primaries were examined through linkage to Osaka Cancer Registry as of the end of December 1995. Observed number of second primaries (0) was compared with the expected according to the incidence in general population (E). Proportional hazards model was used to estimate hazard rate ratio (HR) for second primaries. Results: 149 second primaries were observed. Seventeen cases, detected within 2 months after diagnosis of gastric cancer, were excluded. O/E ratio was 1.13 for adjuvant chemotherapy, 0.93 for immuno-chemotherapy, and 0.78 for immunotherapy, while 1.14 for operation only (not significant). Age-, sex-, and stage-adjusted HR was 1.02 (95% C.I. 0.69-1.50) for chemotherapy, 0.80 (0.41-1.57) for immuno-chemotherapy, and 0.60 (0.08-4.34) for immunotherapy, as compared with the risk for operation only. Among males, HR for current smokers vs. never smokers was 1.82 (1.02-3.26). Conclusions: Adjuvant therapy for gastric cancer did not significantly increase the risk for second primaries, while smoking elevated it. J Epidemiol, 2000 ; 10 : 300304


INTRODUCTION
Survival of gastric cancer patients has increased during the past several decades in Japan.Five-year relative survival was almost 80% for the patients on whom curative operation could be performed ').In order to prevent cancer recurrence and death, adjuvant immuno-chemotherapy after curative resection has been conducted in considerable cases with gastric cancer in Japan2).With the increased survival and the common use of adjuvant therapy, it becomes important to evaluate the risk for second primaries.Recently, Fujimoto 3.4) has issued a stern warning against the adjuvant therapy, presenting his own study results that the immuno-chemotherapy increased the risk for second primaries considerably.His study seems, however, to have serious shortcomings from the epidemiological point of view.In this paper, we evaluate the risk for second primaries among gastric cancer patients and try to verify whether or not adjuvant therapy after gastrectomy increased the risk for second primaries.Associations between smoking and drinking habits at diagnosis and the risk for second primaries were also studied.

SUBJECTS AND METHODS
Study subjects were 1,631 patients who were newly diagnosed with gastric cancer in the stages I, II, III (the General Rules for the Gastric Cancer Study in Surgery and Pathology by the Japanese Research Society for Gastric Cancer in 1985 5)) and underwent curative operation at the Osaka Medical Center for Cancer and Cardiovascular Diseases (OMCC) during the period of 1978-92.They were restricted to patients living in Osaka and aged 15-79 years old at the time of diagnosis.Those who had a past history of cancer were excluded.They were identified from the hospital cancer registry of OMCC, which has collected information on cancer diagnosis, extent of the disease, first course of the treatment, and some lifestyle factors.The analyses were based primarily on this information except for the extent of the disease, since it was used in populationbased cancer registry but not in clinical medicine.Thus we referred to the stage of gastric cancer that was classified according to the General Rules for the Gastric Cancer Study by the Japanese Research Society for Gastric Cancer 5) and was kept in the department of Gastroenterological Surgery of OMCC.
Smoking habits were classified into three categories.Current smokers were persons who smoked tobacco regularly at the time of diagnosis or symptom development.Ex-smokers were those who gave up smoking before the diagnosis or symptom development.Nonsmokers were those who had never smoked regularly.Daily drinkers were persons who drank alcohol almost everyday at the time of diagnosis or symptom development.Ex-daily drinkers were those who stopped daily drinking before the diagnosis or symptom development.Non-daily drinkers were those who had drunk occasionally or never drunk.
Incidence of second primaries was examined through the record linkage with the Osaka Cancer Registry (OCR) 6,7) until the end of December 1995.Their vital status as of the end of December 1995 was confirmed through follow-up survey by the hospital cancer registry of OMCC.The observed number of second primaries (0) was compared with the expected number (E), which was calculated based on the age-, sex-, and calendar year-specific incidence rates of cancer in Osaka and the corresponding person-years of observation.The person-years were accumulated from the date of gastric cancer diagnosis to the date of diagnosis for second primaries, death, and the end of December 1995 whichever came first.A computer program developed by Monson 8) was used for calculations of the person-years, E and 95% confidence interval (C.I.) of O/E ratio.Cox proportional hazards model was used to estimate the hazard rate ratio (HR) for the development of second primaries among the study subjects with adjuvant therapy as compared to the risk among those without it.Associations between smoking and drinking habits and the risk for second primaries were also studied with this model.The SAS/PC statistical package (SAS Institute, Cary, NC) was employed for these analyses.Reported p-values were all two-tailed and those less than 0.05 were considered statistically significant.

RESULTS
During the follow-up period, 149 second primaries were observed among the 1,631 gastric cancer patients.Seventeen cases of those were diagnosed within 2 months after the diagnosis of gastric cancer, so they were excluded from the analyses.Thus the total number of study subjects was 1,614.As of the end of December 1995, 1,064 patients were alive, while 532 were dead.Only 18 patients' vital statuses were unknown.Follow-up periods ranged from 2 months to 17 years and 11 months (mean: 9 years and 3 months).
Table 1 shows the observed number of second primaries, O/E ratio and its 95% C.I. according to the site of second primaries.For all sites except for gastric cancer, the O/E ratio was 1.11 (95%C.I.: 0.93-1.32),not significantly different from 1.0.Distribution of the incidence of second primaries was not significantly different from cancer incidence among the general population, and no specific significant associations were observed with sites of second primaries.
Table 2 presents the number of study subjects, the observed number of second primaries, O/E ratio and its 95% C.I. for all sites except for gastric cancer, according to the type of adjuvant therapy; chemotherapy alone (chemotherapy), both chemotherapy and immunotherapy (immuno-chemotherapy), immunotherapy alone (immunotherapy), and no adjuvant therapy (operation only).The O/E ratio was 1.13 (95% C.I.: 0.82-1.52)for adjuvant chemotherapy, 0.93 (0.45-1.71) for immuno-chemotherapy, and 0.78 (0.01-4.36) for immunotherapy, while it was 1.14 (0.90-1.42) for operation only.All these ratios were not significantly different from 1.0.
Age, sex, and the stage distributions were somewhat different according to the type of adjuvant therapy.In Table 3, therefore, the age-, sex-, and stage-adjusted hazard rate ratio (HR) for second primaries was estimated as compared with the risk among those without adjuvant therapy.The HR was 1 .02(95% C.I.: 0.69-1.50)for chemotherapy, 0.80 (0.41-1.57) for immuno-chemotherapy, and 0.60 (0.08-4.34) for immunotherapy.
Table 4 shows the results on the associations between smoking and drinking habits and the risk for second primaries for males.The HR was 1.82 (1.02-3.26)for current smokers , 0.95 (0.42-2.13) for ex-smokers, as compared with the risk among those who had never smoked after adjustment for age , stage, type of adjuvant therapy and drinking habit .Relation between smoking habits and the risk for second primaries was re-evaluated in terms of smoking-related cancers; that is , the event of the second primaries was restricted to cancers of the esophagus, liver, pancreas, larynx, lung and bladder.Consequently the HR was increased to 3 .00(1.24-7.28)for current smokers, and 1.04 (0.32-3.39) for ex-smokers .Among male current smokers, the elevated O/E ratios more than 1 .5 were observed for colorectum (0:17,E:7 .70),lung (0:17,E:11.30) and bladder (O:3,E:1.98).Although daily drinkers did not show a higher Table 1.Observed number of second primaries, observed-to-expected (O/E) ratio and its 95% confidence interval (95%CL) accordine to nrimarv site.
Table 2. Number of study subjects, observed number of second primaries, observed-to-expected ratio and its 95% confidence interval for all sites except for stomach , according to the type of adjuvant therapy.Table 3. Age-, sex-, and stage-adjusted hazard rate ratio for second primaries and its 95% confidence interval.
Table 4. Smoking and drinking habits and the risk for second primaries for males , risk for second primaries, ex-daily drinkers were suggested to have a higher risk for second primaries , as compared with the risk among non-daily drinkers after adjustment for age , stage, type of adjuvant therapy and smoking habit .Among male exdaily drinkers, an elevated O/E ratio more than 1 .5 was observed only for liver (O:5,E:0.61).Associations between the risk for second primaries and the type of adjuvant therapy did not change appreciably after adjustment for smoking and drinking habits (data not shown).

DISCUSSION
The present study did not show significantly increased risk for second primaries after adjuvant therapy for gastric cancer .This finding is inconsistent with Fujimotos' 3) , in which 858 gastric cancer patients who had undergone gastrectomy during 1963-81 were followed up for second primaries.He reported that those who had undergone adjuvant chemotherapy, and especially, immuno-chemotherapy had significantly increased risk for second primaries on and after 5 years postgastrectomy as compared with the patients who had not (the number of incidence / study subjects: 21/304, 7/64, 21/490, respectively).In comparison with this study, our present study has several strengths.First, identification of second primaries seems to be systematic and more complete because of the use of record linkage with the population-based cancer registry, OCR.Completeness of the registration in the OCR has been fairly good during the last 20 years, although not perfect 6,7).In this study, 41 cases of the second primaries had been diagnosed at other hospitals than OMCC.Second, an external comparison by the O/E ratio was employed, together with an internal comparison by the proportional hazards model.When we refer to other studies' results on the second primaries among gastric cancer patients in Osaka 9.10), which did not show a significant increase of the risk for second primaries as a whole, we suppose that the present study could estimate the risk for second primaries very reasonably.Third, possible confounders such as age, sex and stage of the disease were all adjusted with a multivariate analysis.Last, the sample size of the study subjects was larger than the Fujimotos', and percentage of the lost to followups was very small (1.1%, 18/1,614).Thus it seems that our study results would be more reliable than the Fujimotos' .Although our study did not have enough statistical power to deny the mild or moderate risk for second primaries, the estimated confidence intervals would reject Fujimoto's hypothesis that adjuvant therapy would produce almost twice the risk for second primaries.
Increased risk for second primaries among current smokers and ex-daily drinkers was observed in this study.Smoking is the most important, established risk factor for various cancers and the excess risk among current smokers was observed for sites of the colorectum, lung and bladder in this study .Since the latter two sites are at least smoking related cancers lu , it is most likely that smoking would be causally related to the development of second primaries.No excess risk was found among ex-smokers, which would be a strong motivation for smoking gastric cancer patients to quit smoking.As for drinking, excess risk was only observed for liver cancer among exdaily drinkers.Although there is no evidence supporting it , the incidence of liver cancer would be a consequence of liver disease due to excess alcohol drinking 12).It is unlikely that quitting daily drinking would be a risk factor for second primaries .
Several limitations of this study should be discussed .First, details of the adjuvant therapy were not abstracted nor recorded in the hospital cancer registry.As adjuvant chemotherapy , 5-fruorouracil (5-FU) and/or tegaful (FT) or tegafur uracil (UFT), plus mitomycin C (MMC) were most frequently used in the OMCC during those study periods , while cytarabine (CA), cisplatin (CDDP) and methotrexate (MTX) were used less frequently.As adjuvant immunotherapy, picibanil (OK432), krestin (PSK) and levamisole (LMS) were used often.Commonly used regimens as immuno-chemotherapy were UFT+MMC+OK432, FT+MMC+PSK, or 5FU+MMC+LMS.Lack of information on dose and duration of the medication would be a serious shortcoming of this study.Prospective and systematic recording system for prescribed medicine will be essential for accurate monitoring and surveillance for longterm benefits and side effects of adjuvant therapy.Second, information on smoking and drinking habits was missing for nearly 10% of the male study subjects.Although we included these study subjects for analyses by using a dummy variable, we should be very cautious for interpretation of the study results on associations between the lifestyle factors and the risk for second primaries.Prospective and systematic recording system on lifestyle factors will be also important and mandatory for accurate risk assessment for second primaries.Third, sample size was not enough to get a more reliable conclusion, especially when we evaluate the risk for second primaries according to the type of adjuvant therapy and anti-tumor agent.
In conclusion, adjuvant therapy for gastric cancer was not suggested to significantly increase the risk for second primaries, while smoking habits seemed to elevate the risk for them.Because of several limitations inherent in this study , however, further researches, particularly multicenter co-operative ones will be necessary for exact evaluation, together with establishing information systems on the history of medication in Japan.