Dr. Rajesh P. Dikshit worked on the current study under the tenure of an American Cancer Society Beginning Investigators Award from the International Union Against Cancer.
Risk factors for the development of second primary tumors among men after laryngeal and hypopharyngeal carcinoma
A multicentric European study
Article first published online: 25 APR 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 11, pages 2326–2333, 1 June 2005
How to Cite
Dikshit, R. P., Boffetta, P., Bouchardy, C., Merletti, F., Crosignani, P., Cuchi, T., Ardanaz, E. and Brennan, P. (2005), Risk factors for the development of second primary tumors among men after laryngeal and hypopharyngeal carcinoma. Cancer, 103: 2326–2333. doi: 10.1002/cncr.21051
- Issue published online: 18 MAY 2005
- Article first published online: 25 APR 2005
- Manuscript Accepted: 18 JAN 2005
- Manuscript Revised: 6 DEC 2004
- Manuscript Received: 4 OCT 2004
- Swiss Government
- Ministry for University, Research, Science and Technology (MURST)
- Italian Association for Cancer Research (AIRC)
- Compagnia San Paolo/International Foundation for Research in Experimental Medicine (FIRMS)
- second primary tumors;
- laryngeal carcinoma;
- hypopharyngeal carcinoma;
- standardized incidence ratio;
Second primary tumors (SPT) constitute a major threat to the survival of patients with laryngeal carcinoma. However, to the authors' knowledge little is known regarding the risk factors for developing SPTs or about the strategy to be followed to avoid them.
Eight hundred seventy-six male patients with laryngeal/hypopharyngeal carcinoma enrolled in a population-based, case–control study in 5 centers from South Europe during 1979–1982 were followed up to ascertain the occurrence of SPTs. Standardized incidence ratios were calculated to estimate the risk of SPT occurring in the cohort. Cox proportional hazard models were fitted to estimate the hazard ratio for development of SPTs in relation to use of tobacco smoking, alcohol consumption, and dietary habits before the first primary tumor.
One hundred forty-five patients developed an SPT with an annual average rate of 2.1%. An excess risk of developing an SPT of the tongue, mouth, esophagus, and lung was observed. No elevated risks of SPTs were observed in other organs. Alcohol consumption strongly increased the risk of developing an SPT of the upper aerodigestive tract (UADT). Heavy cumulative cigarette smoking increased the risk of developing a lung SPT. A protective effect of high intake of citrus fruit was noticed for SPT in the lung, whereas high butter intake was associated with an increased risk for SPT of the UADT.
Continuous medical surveillance was essential in the UADT and lung to reduce the risk from an SPT after initial laryngeal/hypopahryngeal carcinoma. Alcohol consumption before the first primary tumor was a risk factor for SPT of the UADT. The study suggested a protective role of citrus fruits in the development of a lung SPT. Cancer 2005. © 2005 American Cancer Society.
It is well known that patients with head and neck carcinomas are at an increased risk of developing second primary tumors (SPT).1–3 However, to our knowledge, little data are available regarding the role of etiologic factors and means of prevention for development of SPT among high-risk patients. The average annual SPT risk of patients with head and neck carcinoma is reported to be 4–6%1 and high levels of tobacco smoking and alcohol consumption are proposed to be important determinants of SPT.4
The current study uses data from a multicentric case–control study on laryngeal and hypopharyngeal carcinoma conducted in southern Europe between 1979 and 1982. We considered a series of > 900 cases with laryngeal/hypopharyngeal carcinoma followed up for 21 years with information on tobacco, alcohol, and dietary habits obtained at the time of diagnosis of the index tumor. We assess the risk of developing SPTs and the role of tobacco, alcohol, and dietary factors in their development. Such information should be useful in the development of strategies aimed at prevention of SPT among high-risk individuals.
MATERIALS AND METHODS
A multicentric population-based, case–control study of laryngeal/hypopharyngeal carcinoma was carried out in six European regions covered by a cancer registry: the canton of Geneva (Switzerland), the department of Calvados (France), the city of Turin and the province of Varese (Italy), and the provinces of Navarra and Zaragoza (Spain). A common protocol was designed and interviews were conducted to define the role of tobacco, alcohol, diet, and occupation in the development of these cancers. The details of study design have been previously reported.5 The patient group enrolled in the study was followed up for the occurrence of SPT by linkage with population, mortality, and cancer registry files. The follow up could not be performed for the cases from Calvados. During the course of follow-up, vital status and cause and date of death were also determined. For each subject, the starting date of follow-up was defined as the date of diagnosis of laryngeal or hypopharyngeal carcinoma. The end of follow-up was set to December 31, 2000, or date of diagnosis of SPT, date of departure from cancer registry area, or date of death if these events occurred earlier. Of the 997 cases with laryngeal/hypopharyngeal carcinoma included in the case–control study, vital status was not known for 25 cases (2.5%). An additional 37 (3.7%) and 7 (0.7%) cases were excluded, as the site of an SPT was that of the first primary tumor and the date of an SPT was not known, respectively. Thus we had complete information for 928 cases (93.1%). Of 928 cases, 52 (5.6%) were females. Because only 6 of them developed SPT, we restricted our analyses to 876 male cases with laryngeal/hypopharyngeal carcinoma.
The criteria for recognition of an SPT were essentially clinical and followed those suggested by the International Agency for Research on Cancer.6 These criteria emphasized that an SPT must be clearly distinct from the first tumor in terms of location but not necessarily in terms of time of appearance. Nonmelanoma skin carcinoma (International Classification of Diseases [ICD9] 173) was not considered to be an SPT.
Person-years at risk were classified by 5-year age groups and by time since entry into the study. The expected number of SPTs was calculated by applying the gender, age, and calendar period-specific cancer incidence rates occurring in each of the five study populations to the person-years at risk and by summing over all age groups and time periods among male patients. The observed number of SPTs was divided by the expected number to obtain the standardized incidence ratio (SIR). The 95% confidence intervals (95% CI) were computed under the assumption of a Poisson distribution for the observed SPTs.
The Cox proportional hazards model was used to estimate the hazard ratio (HR) for SPT and its 95% CI among male patients. The stratified log-rank test was used to test the equality of survivor function across the groups. The models include terms for age, center, socioeconomic group, average amount of alcohol intake in grams per day, average cigarette consumption in cigarettes per day, and site of primary tumor (larynx, hypopharynx). The models used to present results on dietary habits include all the above terms in addition to the term for dietary habits under investigation. The average number of cigarettes smoked per day, alcohol consumption, and dietary factors were categorized as in previous analyses.7, 8 Cumulative cigarette exposure was measured using pack-years. Pack-years of cigarette smoking was categorized into 4 groups using smokers of < 21 pack-years of cigarette smoking as a reference category.
Details regarding tobacco smoking, alcohol consumption, diet history method, and conversion into nutrient intake have been previously described.7, 9 Two broad occupational groups defined occupation status. These groups were chosen to reflect education level and economic status. HR estimates are presented for development of SPT at all cancer sites combined, as well as in the upper aerodigestive tract (UADT; ICD9 140–150) and in the lung (ICD9 162). Analyses were performed using Stata software, release 8.0 (Stata, College Station, TX).10
Of the 876 male subjects included in the study, 145 (16.5%) developed the SPT during a total follow-up period of 6782.8 person-years, resulting in an average annual SPT incidence of 2.1%. The baseline characteristics of the cases with laryngeal/hypopharyngeal carcinoma with and without SPT are displayed in Table 1. No statistically significant difference between patients with and without SPT was observed for age, alcohol consumption, and cigarette smoking. However, there was a significant difference between occupational groups and the centers.
|Characteristics||Cases without SPT (%)||Cases with SPT (%)|
|Age at first cancer (yrs)|
|<50||131 (84.5)||24 (15.5)|
|50–59||236 (82.2)||51 (17.8)|
|60–69||227 (80.8)||54 (19.2)|
|≥70||137 (89.5)||16 (10.5)|
|Turin||250 (92.2)||21 (7.8)|
|Varese||194 (77.6)||56 (22.4)|
|Navarra||109 (85.2)||19 (14.8)|
|Zaragoza||117 (83.6)||23 (16.4)|
|Geneva||61 (70.1)||26 (29.9)|
|Unskilled worker||200 (88.9)||25 (11.1)|
|Skilled worker/professional||523 (81.7)||117 (18.3)|
|Average alcohol drinking (g/day)|
|0–40||174 (88.3)||23 (11.7)|
|41–80||187 (82.4)||40 (17.6)|
|81–120||169 (82.0)||37 (18.0)|
|≥121||201 (81.7)||45 (18.3)|
|Average tobacco smoking (cigarettes/day)|
|0–15||175 (86.2)||28 (13.8)|
|16–25||375 (84.1)||71 (15.9)|
|≥26||176 (80.0)||44 (20.0)|
|Pack-years of cigarette smoking|
|0–20||97 (84.3)||18 (15.7)|
|21–40||298 (82.3)||64 (17.7)|
|41–60||229 (86.7)||35 (13.3)|
|>60||102 (79.7)||26 (20.3)|
Most SPTs (60.4%) were observed in the aerodigestive tract, notably the oral cavity, esophagus, and lung. Table 2 presents the SIR for SPT at selected sites and years of follow-up. The tongue had the highest elevation in risk (SIR, 12.2) followed by the esophagus (SIR, 9.0), mouth (SIR, 7.2), and lung (SIR, 2.6). No increase was observed for colon carcinoma, prostate carcinoma, and bladder carcinoma, or for other cancer sites not presented in Table 2.
|SPT (ICD9)||Time since occurrence of first cancer|
|0–5 yrs||6–10 yrs||>10 yrs||All yrs|
|Obs||Exp||SIR||95% CI||Obs||Exp||SIR||95% CI||Obs||Exp||SIR||95% CI||Obs||Exp||SIR||95% CI|
The HR for developing SPT according to selected risk factors is presented in Table 3. The risk for developing SPT at all sites combined increased with increasing age at first tumor. No trend with age, however, was observed for development of SPT of the UADT. No effect of occupational groups on development of SPT was observed after adjusting for potential confounders. Alcohol affected strongly the development of SPT, in particular UADT carcinomas. Smoking > 60 pack-years was associated with an increased risk of the development of SPT in the lung. No association was found between duration of cigarette smoking and risk for development of SPT in the lung (data not shown). We also examined interactive effects between smoking and alcohol consumption for SPT developing at all sites combined. The effect was consistent with a multiplicative model of interaction, although the number of SPTs at specific sites was too small to adequately evaluate the nature of the smoking–alcohol interactive effect. The risk of developing an SPT was lower for primary laryngeal compared with primary hypopharyngeal carcinoma, although the difference was not significant.
|Variable||Total cases||SPT at all cancer sites combined||SPT of UADT||SPT of lung|
|No.||HR||95% CI||No.||HR||95% CI||No.||HR||95% CI|
|Age at first cancer (yrs)|
|Stratified log-rank test, P value||0.0002||0.96||0.01|
|Average alcohol drinking (g/day)|
|Stratified log-rank test, P value||0.10||0.003||0.06|
|Average cigarette smoking (cigarettes/day)|
|Stratified log-rank test, P value||0.28||0.63||0.09|
|Pack-years of cigarette smokingc|
|Stratified log-rank test, P value||0.23||0.70||0.06|
|Site of first cancer|
Table 4 displays the HR for developing an SPT according to selected dietary items. A high intake of citrus fruits seemed to confer protection for the development of an SPT, particularly in the lung. Increased butter consumption was significantly associated with a higher risk for developing an SPT, particularly in UADT organs. No significant effect for food groups not presented in Table 4, such as meat, poultry, eggs, milk, and olive oil, was observed. We also examined the role of various nutrients in the development of SPT (data not presented). None of the nutrient intake except vitamin C was significantly associated with the development of an SPT. The highest levels of vitamin C intake were inversely related to the development of an SPT at all cancer sites combined (HR = 0.5; 95%CI = 0.3–0.8), and in the lung (HR = 0.3; 95%CI = 0.09–0.7).
|Dietary Intake (g/day)||SPT at all cancer sites combined||SPT of UADT||SPT of lung|
|Total cases||No.||HR||95% CI||No.||HR||95% CI||No.||HR||95% CI|
|Stratified log-rank test, P value||0.34||0.99||0.40|
|Stratified log-rank test, P value||0.32||0.57||0.66|
|Stratified log-rank test, P value||.03||0.73||0.28|
|Stratified log-rank test, P value||0.43||0.99||0.73|
|Stratified log-rank test, P value||0.40||0.41||0.26|
Data from a large multicentric case–control study were used to assess the risk of developing SPTs among patients with laryngeal and hypopharyngeal carcinoma. The multicentric nature of the project entailed many challenges in the standardization of information on various risk factors collected in the study. These problems have been extensively addressed in previous publications.5, 9 A very high SPT ascertainment assured that bias due to differential selection of SPTs during follow-up is unlikely. Further, the information on risk factors was collected using a common protocol. However, we had no information on smoking, alcohol consumption, and other exposures in the interval between the first tumor and the occurrence of SPT. This precluded the quantification of benefits from quitting smoking or drinking after the first diagnosis of cancer is made. As many patients will have reduced their cigarette and smoking consumption, or quit altogether, this will undoubtedly have reduced the association between these factors and an SPT. Another limitation of the study is that we do not have information on TNM classification. If the risk factors being studied were related to the stage of disease (i.e., to disease progression), then it would affect our risk estimates. However, it has been reported that the incidence of SPTs is independent from the primary tumor staging in patients with laryngeal and hypopharyngeal carcinoma.11
The number of SPTs differed between centers. This may be explained by the different clinical stage at diagnosis of the indexed tumor and by differences in background rate. However, the SIRs were similar among the study centers. Further, the selection of the study centers was based on common characteristics of the populations regarding the consumption of a large amount of alcohol, mainly wine, and a common widespread smoking habit in all these southern European countries. The risk for developing an SPT at all sites combined was significantly increased during the entire period of follow-up. In agreement with other studies,12–14 the greatest excess risk was observed for the tongue, mouth, esophagus, and lung. Two alternative hypotheses have been proposed to explain the excess risk for development of SPTs.15 One hypothesis is that, after repeated carcinogenic exposure, the mucosa accumulates genetic alterations that result in the induction of multiple, independent, malignant lesions. An alternative theory for occurrence of multiple tumor states that any transforming event is rare and that the multiple lesions arise due to widespread migration of transformed cells through the entire aerodigestive tract. The first hypothesis refers to the theory of field cancerization as proposed by Slaughter et al.,16 which is now widely accepted and is supported by molecular findings with the demonstration of similar genetic alteration in the adjacent areas.17 In the current study, the excess risk, which is > 10-fold, for tongue and mouth carcinoma was observed during the period of 6–10 years. The risk seems to decrease after this period. For the esophagus, the excess risk was very high up to first 10 years, after which the risk decreased but still was higher (approximately 5-fold). The risk of development of a lung SPT was almost twofold throughout the study period. However, no excess risk was observed for the colon, prostate, bladder, and other sites not presented in Table 2. A significant excess risk for colon carcinoma after primary laryngeal carcinoma was, however, observed in a study based on the Connecticut cancer registry.18 In summary, a high excess risk of SPT was observed for sites where a combined effect of tobacco and alcohol exists (i.e., the oral cavity and oesophagus). A moderate excess risk was observed for sites with a very strong effect of tobacco (i.e., lung carcinoma). For the sites with a moderate association with tobacco (e.g., the bladder and pancreas), no sizable excess in risk was observed for the development of an SPT.
These observations, based on well functioning registries, are unlikely to be explained by surveillance bias or the reporting of recurrent tumors. Further, the effect of treatment (mainly radiotherapy) alone cannot explain the high risk observed as no excess risk was observed for radiosensitive tumors (e.g., the salivary gland) and the excess risk was consistent throughout the study period. It is probable that the excess risk is due to the sharing of common risk factors (e.g., tobacco and alcohol). Given the very high risk of SPT development during the initial years of follow-up, it is plausible that the process of carcinogenesis might have started in more than one organ. Differentiated organ-specific cells from different organs of the body still have many structural and functional characteristics in common. If a carcinogen can cause critical mutations early in carcinogenesis in the cells from one organ, then it is plausible that it could have had a similar effect in cells from different organs.19 Molecular analyses have confirmed the presence of genetically altered fields in an area with histologically normal tissue, including loss of heterozygosity, microsatellite alterations, chromosomal instability, mutations in the Tp53 gene17, 20 as well as markers of high proliferative capacity.21
Similar to findings from other studies,3, 22 increased age at diagnosis of primary tumors was significantly associated with an increased risk of SPT for all sites combined and for the lung. Consumption of alcohol was associated with more than a threefold increase in the risk of development of UADT SPT. Given the large body of evidence regarding alcohol intake and risk of UADT,23 these findings are not surprising. In all the study centers, there was consumption of a large amount of alcohol, mainly wine. This is in contrast with northern Europe where alcohol consumption is lower and beer and liquor consumption is predominant. The lack of heterogeneity precluded an analysis by type of alcoholic beverage. A strong increase in the risk of developing a lung SPT was observed for heavy cumulative cigarette smoking. However, no significant association was observed between the risk of development of an SPT in the lung and the average number of cigarettes smoked per day, and duration of cigarette smoking. This could be explained because of homogeneity of the study cohort in relation to smoking behavior. As most cases were heavy smokers, no proper reference category of nonsmokers/weak smokers could be used. Further, some of the cases with primary laryngeal carcinoma might have stopped smoking after development of the first primary tumor, resulting in a decrease in the risk related to smoking habits. Assuming that the misclassification of postdiagnostic smoking habits is nondifferential with respect to the development of an SPT, we have probably underestimated the effect of smoking on SPT risk. When the analysis was restricted to cases with laryngeal carcinoma only (excluding hypopharyngeal carcinoma), the risk of developing a lung SPT with the highest level of cigarette consumption was observed to be almost 3-fold (odds ratio =2.7; 95%CI, 1.0–7.3). It has been shown that laryngeal carcinoma is more related to smoking habits whereas hypopharyngeal carcinoma is more related to the heavy consumption of alcohol.7 Few other studies have similarly observed a nonsignificant increase for the development of an SPT in relation to the heavy consumption of tobacco,24, 25 and one study observed a decrease in the risk of development of SPT after quitting the tobacco habits.26
Similarities and differences in dietary habits across the study centers have been previously discussed.7 An increased risk observed for the development of SPT in the UADT with the heavy consumption of butter is plausible given the possible role of dairy products in the development of head and neck carcinoma.27 Protection conferred by the heavy intake of citrus fruit and to some extent by vitamin C in particular against the development of SPT in the lungs is in accordance with the protective role of this group of foods in the development of lung carcinomas.28 However, the effect of dietary items on SPT development should be interpreted with caution. We could not exclude the possibility that the observed association is due to residual confounding of socioeconomic and behavioral factors especially as social and behavior factors have the potential to strongly confound associations between vitamins and disease.29 We could not exclude the possibility of false-positive findings due to multiple comparisons performed to observe the effects of dietary factors on risk of development of SPTs. Further, as discussed, we do not have dietary histories after the diagnosis of first primary tumors, which could also affect our estimates.
In summary, we conclude that there is an excess risk for the development of an SPT in the UADT and lung after initial laryngeal/hypopharyngeal carcinoma but no excess risk was observed in organs outside the respiratory tract and UADT. It would be important to continuously monitor the patients with primary laryngeal/hypopharyngeal carcinoma for a long term. Stopping tobacco smoking and alcohol consumption as well as increasing consumption of citrus fruits should be recommended to patients with laryngeal/hypopharyngeal carcinoma as a mean to reduce the risk of SPTs.
The authors acknowledge the contribution of Drs. J. Esteve, E. Riboli, G. Pequignot, L. Raymond, B. Terracini, and F. Berrino and the late Drs. A. Tuyns and W. Lehmann in the design and conduct of the original case–control study. They also acknowledge the contribution of Drs. D. Mirabelli, S. Rosso, R. Zanetti, M. Chiusolo, and L. Nonnato.
- 1Upper aerodigestive tract cancers. In NeugutA, MeadowsA, RobinsonE, editors. Multiple primary cancers. Philadelphia: Lippincott Williams & Wilkins, 1999: 319–346., , .
- 4Multiple primary cancers as clues to environmental and heritable causes of cancer and mechanism of carcinogenesis. IARC Sci Publ. 2004;(157): 289–297., .
- 6Multiple primary cancer. In HigginsonJ, MuirCS, MunozN, editors. Human cancer: epidemiology and environmental causes. Cambridge: Cambridge University Press, 1992: 493–496., , .
- 9Cancer of the larynx/hypopharynx, tobacco and alcohol: IARC international case-control study in Turin and Varese (Italy), Zaragoza and Navarra (Spain), Geneva (Switzerland) and Calvados (France). Int J Cancer. 1988; 41: 483–491., , , et al.
- 10STATA Corporation. Stata statistical software: release 8.0. College Station, TX: Stata Corporation, 2003.
- 12Second cancer occurring after cancers of the mouth and pharynx: data from three population based registries in Australia, Scotland and Slovenia. Oral Oncol. 1995; 31: 315–318., , , , .
- 18Second cancer following cancer of the respiratory system in Connecticut, 1935-1982. Natl Cancer Inst Monogr. 1985; 68: 83–98.,
- 19Common environmental risk factors. In NeugutA, MeadowsA, RobinsonE, editors. Multiple primary cancers. Philadelphia: Lippincott Williams & Wilkins, 1999: 213–223., .
- 23International Agency for Research on Cancer monographs on the evaluation of carcinogenic risk to humans: alcohol drinking. Volume 44. Lyon: International Agency for Research on Cancer, 1988.
- 28International Agency for Research on Cancer handbooks of cancer prevention. Volume 8. Fruit and vegetables. Lyon: International Agency for Research on Cancer, 2003.