Second primary tumors after prostate carcinoma

Authors

  • Fabio Levi M.D.,

    Corresponding author
    1. Registre Vaudois des Tumeurs, Institut Universitaire de Médecine Sociale et Préventive, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
    2. Registre neuchâtelois des tumeurs, Neuchâtel, Switzerland
    • Registre vaudois des tumeurs, Institut Universitaire de Médecine Sociale et Préventive, Centre Hospitalier Universitaire Vaudois, Falaises 1, Casier 15, CH-1011 Lausanne, Switzerland
    Search for more papers by this author
  • Lalao Randimbison Sc.D.,

    1. Registre Vaudois des Tumeurs, Institut Universitaire de Médecine Sociale et Préventive, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
    Search for more papers by this author
  • Van-Cong Te M.D.,

    1. Registre Vaudois des Tumeurs, Institut Universitaire de Médecine Sociale et Préventive, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
    Search for more papers by this author
  • Georges Erler M.D.,

    1. Registre neuchâtelois des tumeurs, Neuchâtel, Switzerland
    Search for more papers by this author
  • Carlo La Vecchia M.D.

    1. Istituto di Ricerche Farmacologiche “Mario Negri,” and Istituto di Statistica Medica e Biometria, Università degli Studi di Milano, Milan, Italy
    Search for more papers by this author

Abstract

BACKGROUND

Several large datasets have shown a reduced risk of all neoplasms after a diagnosis of prostate carcinoma but an increased incidence rate of urologic carcinoma has been suggested.

METHODS

Data collected by the Cancer Registries of the Swiss Cantons of Vaud and Neuchâtel (approximately 760,000 inhabitants) were used to estimate the incidence rate of a second primary tumor after a diagnosis of prostate carcinoma. A total of 4503 cases registered between 1974 and 1994 were followed until the end of 1996 (17,065 person-years).

RESULTS

A total of 380 second primary neoplasms were observed versus 534.1 expected primary neoplasms (standardized incidence ratio [SIR] = 0.7; 95% confidence interval, 0.6–0.8). SIRs were significantly below unity for lung carcinoma (SIR = 0.7) and other major tobacco-related neoplasms, including those of the mouth or pharynx (SIR = 0.5), esophagus (SIR = 0.4), pancreas (SIR = 0.5), and larynx (SIR = 0.8). There was no excess rate of subsequent urologic carcinoma (SIR = 1.0) or colorectal carcinoma (SIR = 0.9). The reduced SIRs for lung carcinoma were stronger in elderly men (age ≥ 75 years) and in patients with a shorter period since diagnosis (< 5 years).

CONCLUSIONS

The incidence of all neoplasms was reduced significantly in men diagnosed with prostate carcinoma. Selection of the population, underregistration of second primary tumors, and reduced surveillance in elderly men with prostate carcinoma may, at least in part, explain this reduction in risk. No excess risk was observed for the complex of urologic neoplasms nor for tobacco-related neoplasms. This finding would not support an association between cigarette smoking and prostate carcinoma. Cancer 1999;86:1567–70. © 1999 American Cancer Society.

Mortality rates from prostate carcinoma in Switzerland (22.5 per 100,000 in 1990–1994) are highest in the world (i.e., approximately 30% higher than in the U.S.).1 Incidence rates from the Cancer Registries of the Cantons of Vaud and Neuchâtel also were high on a worldwide and European scale (45.7 per 100,000 in Vaud, and 42.3 per 100,000 in Neuchâtel, world standard, 1988–1992). Age-standardized incidence rates increased from 33.1 to 48.6 per 100,000 (+47%). Mortality was relatively stable (20.4–21.0). Five-year relative survival rates increased from 46% to 58%, which is compatible with a finding of improved diagnosis of prostate carcinoma over the last few years.2, 3

Data regarding second primary tumors after prostate carcinoma are open to discussion. Some of the largest datasets, including the Danish Cancer Registry for 1943–1980,4 the Connecticut Cancer Registry for 1935–1982,5 the Finnish Cancer Registry for 1953–1979,6 and the New South Wales (Australia) Cancer Registry for 1972–1991,7 as well as a hospital-based series from Japan8 showed an overall reduced risk of 15–30% for all neoplasms after a diagnosis of prostate carcinoma. There is no obvious explanation for such a reduced incidence rate, apart from selective mechanisms or the possibility of systematic underregistration of other primary neoplasms after a diagnosis of prostate carcinoma. However, this is difficult to quantify.

However, other studies, including one based on the Detroit Metropolitan Area for 1973–1982,9 showed observed numbers of second primary tumors similar to the expected numbers. In addition, there are suggestions that incidence rates of bladder carcinoma and kidney carcinoma may be increased after a diagnosis of prostate carcinoma,7, 10 although a potential ascertainment bias within the complex of urologic neoplasms remains possible.

To provide further information regarding these open issues, we therefore have reanalyzed data regarding second primary tumors after a diagnosis of prostate carcinoma using the Vaud and Neuchâtel Cancer Registry datasets for the period 1974–1994. Data published up to 1989 for the Vaud Cancer Registry showed standardized incidence ratios (SIR) of 0.7, based on 91 observed cases, for all cancer sites, and SIRs of 1.5, based on 11 cases, for infiltrating bladder carcinoma. None of these estimates was significant.11

MATERIALS AND METHODS

Data for the current report were abstracted from the Vaud and Neuchâtel Cancer Registry files, which include incident cases of malignant neoplasms diagnosed in these cantons,12, 13 whose populations were approximately 600,000 and 160,000, respectively, in 1990.14 In these cantons, cancer registration systems have been in place since 1972 and population-based incidence data have been available since 1974. The registries are tumor-based, and multiple primary malignancies found in the same person are entered separately. The basic information available is comprised of sociodemographic characteristics of the patient (age and gender), the primary site and histologic type of the tumor according to the standard International Classification of Diseases for Oncology, Ninth Revision, (ICD-O-9),15 and the date of diagnostic confirmation. Overall histologic confirmation and completeness exceeds 90%.12, 13

Both passive (through computer linkage with official mortality data files, including national and cantonal death/vital statistics data) and active follow-up (through verification of the vital status of apparently nondeceased individuals using registries of current residence) are conducted, and each subsequent item of information concerning an already registered case is used to complete the record of that patient. The proportion of cases lost to follow-up, due to migration, or due to other causes was approximately 1%.

After the exclusion of 502 cases (10%) diagnosed at autopsy (n = 268) or at death (n = 234), the current series was comprised of a total of 4503 cases of first diagnosed prostate carcinoma registered between 1974 and 1994 (mean age, 74.2 years; range, 42–97 years). Histologic confirmation was available for 89.7% of the cases. No information was collected regarding stage of the disease.

Calculation of expected numbers was based on single age specific and calendar year specific incidence rates of all neoplasms and each specific cancer site multiplied by the observed number of person-years at risk. Second primary tumors detected at death or autopsy were included in both observed and expected values. The end of follow-up was determined by a second primary tumor, death, emigration, or the end of the study period (December 31, 1996) for a total of 17,065 person-years at risk. The significance of the observed/expected ratios (SIRs) and the corresponding 95% confidence intervals (95% CI) were based on the Poisson distribution.16

RESULTS

Table 1 gives the observed and expected numbers of various disease sites after a diagnosis of prostate carcinoma. Overall, 380 second neoplasms were observed versus 534.1 expected, corresponding to a SIR of 0.7 (95% CI, 0.6–0.8); excluding nonmelanomatous skin neoplasms, 232 neoplasms were observed versus 398.7 expected (SIR = 0.6; 95% CI, 0.5–0.7). SIRs were significantly below unity for lung carcinoma (SIR = 0.7; 95% CI, 0.5–1.0), and also were below unity for other major tobacco-related neoplasms, including carcinoma of the mouth or pharynx (SIR = 0.5), esophagus (SIR = 0.4), pancreas (SIR = 0.5), and larynx (SIR = 0.8). The SIR also was significantly below unity for liver carcinoma (SIR = 0.3; 95% CI, 0.1–1.0) and was 0.9, based on 50 cases, for colorectal carcinoma. A total of 31 cases were observed for bladder carcinoma versus 24.9 expected (SIR = 1.2; 95% CI, 0.8–1.8); 5 cases of carcinoma of the kidney were observed versus 9.9 expected (SIR = 0.5; 95% CI, 0.1–1.2). None of the other sites showed any consistent pattern, with some showing SIRs slightly above unity and the majority of the others slightly below unity.

Table 1. Observed and Expected Cases and Standardized Incidence Ratios with Corresponding 95% Confidence Intervals of Selected Subsequent Neoplasms after An Initial Diagnosis of Prostate Carcinoma in Vaud and Neuchâtel, Switzerland, 1974–1994
SiteICD-O-9OESIR (95% CI)
  1. ICD-O-9: International Classification of Diseases-Oncology-9th Revision; O: observed; E: expected; SIR: standardized incidence ratios; 95% CI: 95% confidence interval.

Mouth or pharynx140–9713.80.5 (0.2–1.1)
Esophagus15049.70.4 (0.1–1.1)
Stomach1512520.51.2 (0.8–1.8)
Colorectum153–45057.40.9 (0.6–1.1)
Liver15539.00.3 (0.1–1.0)
Gallbladder15633.50.9 (0.2–2.5)
Pancreas157612.20.5 (0.2–1.1)
Larynx16144.90.8 (0.2–2.1)
Lung1624865.20.7 (0.5–1.0)
Skin, melanoma17258.00.6 (0.2–1.5)
Skin, nonmelanoma173148135.31.1 (0.9–1.3)
Bladder1883124.91.2 (0.8–1.8)
Kidney18959.90.5 (0.1–1.2)
Brain19122.90.7 (0.1–2.4)
Non-Hodgkin lymphomas200, 20289.10.9 (0.4–1.7)
Myeloma20334.80.6 (0.1–1.8)
Leukemias204–889.30.9 (0.4–1.7)
Total, all sites140–208380534.10.7 (0.6–0.8)
All sites, minus skin nonmelanoma232398.70.6 (0.5–0.7)

When separate strata of age (< 75 years and ≥ 75 years), time since diagnosis (< 2 years, 2–4 years and ≥ 5 years), and calendar period (Table 2) were considered, the reduced SIRs for lung carcinoma were observed in elderly men and in men with a shorter period since diagnosis. No consistent pattern was observed with reference to calendar period for any other site, nor for total cancer mortality (SIR = 0.8 for age < 75 years and 0.6 for age ≥ 75 years; 0.7 for period since diagnosis of < 5 years and 0.8 for period since diagnosis of ≥ 5 years).

Table 2. Observed Cases and Standardized Incidence Ratios with Corresponding 95% Confidence Intervals of Selected Subsequent Neoplasms after An Initial Diagnosis of Prostate Carcinoma in Strata of Age, Time since Diagnosis, and Calendar Period, (Vaud and Neuchâtel, Switzerland, 1974–1994)
 Age (yrs)Time since diagnosis (yrs)Calendar period (yrs)
< 75≥ 75< 22–4≥ 51974–19841985–1994
OSIR (95% CI)OSIR (95% CI)OSIR (95% CI)OSIR (95% CI)OSIR (95% CI)OSIR (95% CI)OSIR (95% CI)
  1. O: observed cases; SIR: standardized incidence ratios; 95% CI: 95% confidence interval.

Mouth or pharynx40.530.630.520.420.740.630.4
(0.1–1.2)(0.1–1.7)(0.1–1.5)(0.1–1.5)(0.1–2.4)(0.2–1.6)(0.1–1.2)
Colorectum291.1210.7170.7201.0130.9210.8290.9
(0.7–1.5)(0.4–1.1)(0.4–1.2)(0.6–1.6)(0.5–1.5)(0.5–1.2)(0.5–1.9)
Lung360.9120.5110.4190.8181.2220.8260.7
(0.6–1.2)(0.3–0.9)(0.2–0.7)(0.5–1.3)(0.7–1.9)(0.5–1.1)(0.5–1.1)
Bladder181.5131.080.8121.4111.7151.4161.1
(0.9–2.3)(0.5–1.7)(0.3–1.6)(0.7–2.5)(0.6–3.0)(0.8–2.3)(0.6–1.8)
Total, all sites2110.81690.61350.61370.71080.81640.72160.7
(0.7–0.9)(0.5–0.7)(0.5–0.7)(0.6–0.9)(0.6–0.9)(0.6–0.8)(0.6–0.8)
All sites, minus skin nonmelanoma1340.6980.5770.5880.6670.71150.61170.5
(0.5–0.8)(0.4–0.6)(0.4–0.6)(0.5–0.7)(0.5–0.9)(0.5–0.7)(0.4–0.6)

DISCUSSION

The current analysis further supports, on the basis of a more than doubled number of observed cases, that the incidence rate of all neoplasms is reduced significantly in men diagnosed with prostate carcinoma.11 The observation that SIRs for carcinomas of both the lung and liver (i.e., two common sites for metastases) were below unity suggests that underregistration (due to misclassification between primary neoplasms and metastases) may partly explain this reduced risk, particularly in cases with a short time since the diagnosis of prostate carcinoma. However, this does not explain the reduced SIRs for several other sites, which are not common sites of metastases.

More likely, it is possible that medical surveillance is reduced in elderly men with prostate carcinoma.7, 17 This may apply to a larger degree to elderly men with advanced disease, although we had no systematic information regarding the stage of disease. Another potential explanation is that cases of carcinoma of the prostate are diagnosed from a relatively healthy population who take advantage of screening for prostate carcinoma,18 and have other healthy behaviors as well, including not smoking. However, we had no direct information regarding screening procedures. However, going against a major role for such a general surveillance bias is the observation that the number of observed cases of nonmelanomatous skin neoplasms was close to that of expected cases.

With reference to other possible sources of bias, both passive (using cantonal and national death certification data) and active follow-up were adopted, and losses to follow-up, including migration, were restricted to < 1% of cases. Therefore, this is unlikely to have introduced any material bias in the estimates presented.

In any case, it is clear that the incidence rate for the complex of tobacco-related neoplasms was not increased after a diagnosis of prostate carcinoma. Although some studies suggested that cigarette smoking may be associated with the development of prostate carcinoma,19, 20 the absence of excess risk in this dataset confirms the findings of the majority of recent studies on this topic.21, 22

In this study, there was no excess of subsequent urologic malignancies, with 36 observed cases versus 34.8 expected cases. The SIRs were slightly above unity for (infiltrating) bladder carcinoma and below unity for carcinoma of the kidney, although in the absence of any significant or consistent pattern. Likewise, no excess risk was observed for colorectal carcinoma, which has been related to being overweight and consuming a diet rich in (saturated) fats, i.e., potential risk factors for prostate carcinoma.22–24

Thus, the main clinical and public health message of this study is the confirmation of the absence of a noticeable excess of cancer risk after a diagnosis of prostate carcinoma.

Ancillary