Second primary tumours in more than 2-year disease-free survivors of small-cell lung cancer in Japan: the role of smoking cessation.

Patients with small-cell lung cancer who survive more than 2 years have a significantly increased risk (relative risk of 3.6) of developing a second primary tumour. The cessation of cigarette smoking after successful therapy is associated with a significantly decreased risk of a second primary tumour.


MATERIALS AND METHODS Patients
From Januarv 1978 to December 1992. 980 consecutix-e patients with histologically confirmed. previously untreated SCLC were treated at the National Kinki Central Hospital and Osaka Prefectural Habikino Hospital A ith combination chemotherapy with or w-ithout chest radiotherapy.

Definitions
The upper aerodigestive tract includes the epithelial regions of the head and neck. lunc and oesophagus. Smoking-related cancers include cancers of the lung. larvmx. oral cavity including phanrnx. oesophagus. pancreas. bladder. kidney. stomach and uterine cernix (Blum. 1993). Smokinc historin 2-vear cancer-free sunviVors >-as determined by interviewmig those patients still alive at the time of manuscript preparation or the relatives of deceased patients. Smoking cessation was defined as completely stopping smokinLxwithin 6 months after initiation of treatment. The period of the study w-as taken as starting from the first day of chemotherapy administration. and the date of relapse or second Received 7 October 1997Revised 2 February 1998Accepted 5 February 1998 Correspondence to: M Kawahara pnmarn cancer >-as taken as the day of histological or cytological documentation of redevelopment of cancer. The appearance of SCLC more than 2 y-ears after the initiation of therapy A-as defined as relapse.

Statistical analysis
For estimation of the expected x-alues of second cancer desvelopment. the period of risk began 2 years after initiation of treatment for SCLC and ended with the date of death. date of last follow -up or date of diagnosis of a second cancer. w-hichever occurred first.
Age. sex. and period-specific rates for cancer incidence x-ithin the period 196-3-92 were applied to the appropriate person-years of obsernation (Osaka Prefectural Department of Environment andPublic Health. 1993: Osaka Prefectural Department of Health. 1995). The cancer incidence data for 1992 A-ere applied to the person-years from 1992 to 1995. Statistical methods for risk estimation Nvere based on the assumption that the obserned number of second cancers follow-ed a Poisson distribution (Boice et al. 1991 ). To calculate excess risks per 10 000 patients per y ear in subgroups w-ith sianificant relatix-e risks. the expected number of cases w-as subtracted from the number observed. The difference A as divided by person-years of observation. then multiplied by IO-. The risk of a SPT with a specific exposure (e.g. smoking and treatment) x-as estimated by comparing the patients w ithout the specific exposure. using Poisson regression methods (SAS Institute. 1989) adjusting for sex. age (> 65 vs < 65 nears old). performance status (PS) (0-1 Xs 2-4). etoposide. radiotherapy and cumulatiVe smoking amount before diacnosis of SCLC (> 45 pack-y ear vs < 45 pack-years ).

RESULTS
Of the 980 patients of SCLC treated in the two hospitals. 70 (7%5 x-ere alive and free of cancer at least 2 vears after the initiation of treatment. The median survival time of these 70 patients >-as 9.0 years from initiation of treatment for SCLC. Fiveand 10-vear survival rates A-ere 83=/e and 43% respectixely. Ten patients wxere The median follow--up from initiation of therapy was 6.7 vears (ranae' .1-15.1 vears).
Fifteen of the 70 disease-free survivors dexeloped one or more SPTs 2.2-8.1 years (median. 6.1 years) after beginning therapy for SCLC. Details of these patients are shown in Table 1. Five patients (cases 3. 4. 9. 13 and 15) developed a second primary lung cancer (other than SCLC) (four squamous. one adenocarcinoma). of which four occurred in different lobes from the original SCLC. Four of the patients receixved radiotherapy. Two second primarx lung cancers developed outside the radiation field (cases 4 and 9). The other malignancies consisted of carcinomas of the stomach. oesophagus. larynx. prostate. gallbladder and bladder. and acute myelogenous leukaemia.
Of the 70 patients. nine relapsed with SCLC. These relapses occurred 2.0-8.5 years after the initiation of SCLC. Twentx -fixve patients have died: five from recurrent SCLC. 11 from a SPT. The other causes of deaths unrelated to cancer were pulmonary disease (n = 3). cardiac disease (n = 2). cerebrovascular disease (n = 1). neurological disease (dementia after whole brain irradiation for SCLC) (n = 1) and unknown (n=). Table 2 show s the relative and absolute risks of SPT after initiation of therapy for SCLC. The risk for development of any SPT increased significantly by 3.6 [95%7c confidence interval (CI) 2.0-5.9]. This oxerall increase in risk was mainlv due to the 7.0fold increase in lung cancer other than SCLC (95c7% CI 2.3-16.4). a 41.1-fold increase in carcinoma of the larvnx (95%7c CI 4.5-144.4) and a 15.6-fold increase in carcinoma of the oesophagus (95%7c CI 1.7-55.5). The relati-e risk of all upper aerodigestixe cancers (nine patients) was 9.3 (95c CI 4.3-17.7). When smoking-related cancers (12 patients) are combined. the relatixe risk was 5.2 (95c% CI 2.7-9.1).
Smoking status after the initial primary tumour was axailable for all but one patient (case 3 in Table 1). Smoking status was obtained directly from 44 patients (63%). from relatives for 20 (29%) and from the patients' medical records for five (77c). There has been no SPT among the fixve nexver smokers. After initiating therapy for SCLC. 33 patients (49%/c) continued to smoke and 31 patients (48%7c) stopped smoking (Table 3). Of the patients xho continued to smoke. 11 (33%7c) dex eloped a SPT. Of the 31 patients x ho stopped smoking after therapy. only three (10%c) had a subsequent SPT (cancer of the stomach. oesophagus and gallbladder respectixely. see Table 1). Among those wxho continued to smoke. the risk for a SPIT was significantly increased (5.4 times: 95%7c CI 2.7-9.7). relatixe to the aeneral population. In contrast. those wxho stopped smoking showxed only a 1.6-fold increase (95%7c CI 0.3-4.6). which wxas not significantlx different from the lexel in the general population. The relative risk for non-SCLC was sianificantlv increased 12.8-fold (95c/c CI 3.4-32.8) in continuing smokers. No second non-SCLCs have been found amona those x ho stopped smoking.
We assessed the relationship betxxeen continued smokin2 habits and the risk of a SPT. adjusted for sex. PS. age. etoposide treatment. radiotherapy and cumulatixe smoking history. The results are showxn in Table 4. The 33 patients who continued to smoke had a significantly increased risk of a SPF (4.3. 95%7c CI 1.1-15.9. P=0.03). The other factors such as sex. PS. age. cumulative smokine amount. use of the anticancer drug etoposide or radiation had no effect on the dex elopment of a SPI. We assessed the interaction between smoking habits and radiotherapy on the risk of a SPT. Relative to the risk of SPT in patients without previous radiotherapy wxho stopped smokingy. the risk is 0.92 in patients without radiotherapy wxho continued smoking. 0.37 in patients with radiotherapy who stopped smoking. and 2.33 in patients with radiotherapy who continued smoking. The risk of current smoking in patients with previous radiotherapy is 6.30 relatixve to those xith radiotherapy wxho stopped smoking. although this interaction is not statistically significant (P = 0.24). possibly because of the small number of patients.

DISCUSSION
In our study. 15 patients out of 70 long-term surxixors of SCLC had a SPT. The relatixe risk for any SPF compared xxith the Second primary tumour and smoking in SCLC 411 Table 2 Risk of a second primary tumour in patients surviving 2 or more years free of small-cell lung cancer  general population was significantly increased s%ith a relative risk of 3.6 (95%c CI 2.0-5.9). The risk was substantially higher for tumours located in the upper aerodigestive tract and for the total related to smoking. Richardson et al (1993) also report that a risk for any SPT is 4.4 (95% Cl 2.5-7.2).
Smokinc history after treatment of SCLC influenced the risk of development of a SPT. The 33 patients who continued to smoke had a significantly increased risk for a SPF (4.3. 95%c CI 1.1-15.9. P = 0.03) compared with those w ho stopped smoking. Richardson et al ( 1993) reported that the patients who continued to smoke had a threefold increased risk for a second primary luna cancer compared w-ith the patients who stopped smoking. However. w-e could not calculate the risk of second primarv lung cancer because no second primary non-SCLC has been found in those w ho stopped smoking in our patients. There appeared to be an interaction between smoking and chest irradiation. the risk of current smokingr combined w ith previous radiotherapy being 6.30 relative to those with radiotherapy who stopped smokingy. although this interaction is not statistically significant (P = 0.24). perhaps because of the small number of patients. This suggest that. althouah irradiation itself is beneficial for lona-term survivors of SCLC. current smokina after previous irradiation is harmrful to these patients. Recently. Tucker et al (1997) reported a similar interaction for the risk of a second lung, cancer betmeen smoking, after treatment and previous chest irradiation. although. as in the present study. the interaction w-as not statistically significant.
In conclusion. these data indicate that patients swith SCLC who surnive cancer-free for more than ' -ears have a significantly increased risk of developing a SPT. and that the cessation of cigarette smoking, after successful therapy is associated with a decreased risk for a SPT. These data warrant cessation of smokina amonc patients with SCLC and the importance for developing efficient proggrammes to support patients attempting to give up smoking. Cl. confidence interval.