Incidence of Cancer Among Patients with Hidradenitis Suppurativa
Total Page:16
File Type:pdf, Size:1020Kb
STUDY Incidence of Cancer Among Patients With Hidradenitis Suppurativa Jan Lapins, MD; Weimin Ye, MD; Olof Nyre´n, MD; Lennart Emtestam, MD Background: On the basis of some case reports, a re- Results: The risk of developing any cancer in the lationship has been suggested between hidradenitis cohort with HS increased 50% (95% confidence interval suppurativa (HS) and the development of nonmela- of SIR, 1.1-1.8, based on 73 observed cases). Statisti- noma skin cancer. cally significant risk elevations were observed for non- melanoma skin cancer (5 cases; SIR, 4.6; 95% confi- Objectives: To confirm this relationship and to ex- dence interval, 1.5-10.7), buccal cancer (5 cases; SIR, plore the risk of other cancers among patients with HS. 5.5; 95% confidence interval, 1.8-12.9), and primary liver cancer (3 cases; SIR, 10.0; 95% confidence inter- Patients: Patients with a discharge diagnosis of HS were val, 2.1-29.2). obtained from the computerized database of hospital dis- charge diagnoses from January 1, 1965, through December Conclusions: This study confirms an increased risk of 31, 1997. A total of 2119 patients with HS were identified. nonmelanoma skin cancer among patients with HS. The risk for buccal cancer and primary liver cancer was also Setting: All hospitals in Sweden. elevated among this cohort, but these associations should be interpreted cautiously because the combination of mul- Design: With record linkage to the Swedish National tiple significance testing and the few observed cases may Cancer Registry, standardized incidence ratios (SIR [the have generated chance findings. ratio of the observed to expected incidence]) were cal- culated to estimate relative risk. Arch Dermatol. 2001;137:730-734 IDRADENITIS suppurativa 29 among men). The average age at diag- (HS) is a chronic, suppu- nosis of cancer was 51.2 years for women rative, and cicatricial in- and 55.0 years for men (Table 1). Among flammatory disease, all 2119 patients, 134 (6%) had a diagno- mainly affecting apo- sis of alcoholism and 152 (7%) had a di- Hcrine gland–bearing areas of the skin.1 Sev- agnosis of diabetes mellitus. About one eral case reports of co-occurrence of HS third (652) of the patients had ever un- and nonmelanona skin cancer2-16 have im- dergone surgical excision of the affected plied a causal relationship, but firm epi- skin areas. demiological data are lacking. In an at- After excluding 8 cancer cases accu- tempt to confirm this association and to mulated during the first year of fol- look for other cancer associations as well, low-up (SIR, 1.9; 95% CI, 0.8-3.8), in- we performed a retrospective cohort study cluding 1 case of squamous cell carcinoma of patients hospitalized for HS in Swe- of the skin, the risk for any cancer (all sites) den. The patients were followed up for up was increased by 50% among patients hos- From the Department of to 32 years to examine the subsequent risks pitalized for HS, compared with the age- Medicine, Section of of cancer. and sex-matched general Swedish popu- Dermatology and Venereology, lation during the time of the study Karolinska Institute at RESULTS (Table 2). The estimated relative risks for Huddinge University Hospital, cancers of different sites with at least 2 ob- Stockholm, Sweden (Drs Lapins and Emtestam); and the On average, the patients with HS were fol- served cases are also listed in Table 2. Department of Medical lowed up for 9.8 years, yielding 20801 ac- There was a significantly increased rela- Epidemiology, Karolinska cumulated patient-years at risk. During the tive risk for nonmelanoma skin cancer Institute, Stockholm (Drs Ye observation period, we ascertained a to- (5 cases; SIR, 4.6; 95% CI, 1.5-10.7), but and Nyre´n). tal of 81 cases of cancer (52 among women, no excess risk was observed for mela- (REPRINTED) ARCH DERMATOL / VOL 137, JUNE 2001 WWW.ARCHDERMATOL.COM 730 ©2001 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/30/2021 PATIENTS AND METHODS or incomplete national registration numbers. We also ex- cluded 67 patients with prevalent cancers (ie, those who had a previously registered diagnosis of cancer) and 51 pa- STUDY COHORT tients with inconsistencies uncovered during medical record linkage. Thus, a total of 2119 patients, 1495 women In 1964-1965, the National Board of Health and Welfare and 624 men, were enrolled in the study cohort, which is started collecting data on individual hospital discharges in further characterized in Table 1. the Inpatient Register. In addition to the national registra- tion number (unique personal identifiers assigned to all STATISTICAL ANALYSIS Swedish residents), each record—corresponding to one in- hospital episode—contains administrative and medical data Follow-up time (person-years) was calculated from the date such as hospital department and discharge diagnoses. The of enrollment in the cohort (date of the first discharge from diagnoses are coded according to the International Classi- the hospital having a diagnosis of HS) until the occur- fication of Diseases, 7th Revision (ICD-7) through 1968, the rence of a first cancer diagnosis, emigration, death, or the eighth revision until 1987 (ICD-8), the ninth revision end of the study (December 31, 1997). To avoid possible (ICD-9) until 1996, and the 10th revision (ICD-10) there- ascertainment bias associated with differential autopsy rates after. The number of hospitals providing data to the reg- between the cohort members and the general population, ister has increased steadily: the register covered 60% of the we did not count cancers found incidentally at autopsy. Rela- Swedish population in 1969, 75% in 1978, and 85% by the tive risk of cancer was estimated as the standardized inci- end of 1983. From 1987, the register attained complete na- dence ratio (SIR), defined as the ratio of the observed num- tionwide coverage. ber of cancers to that expected. The expected number of All patients recorded in the Inpatient Register with a cancers was calculated by multiplying the number of ob- discharge diagnosis of HS (ICD-7=714.07, ICD-8=705.91, served person-years, divided into age- (in 5-year groups), ICD-9=705W, and ICD-10=L732) were initially selected for sex-, and calendar year–specific strata, by the correspond- inclusion in the study. A total of 2259 unique national reg- ing cancer incidence rates. These incidence rates, derived istration numbers were registered at least once with this di- from the relevant strata in the entire Swedish population agnosis between January 1, 1965, and December 31, 1997. and aggregated by 5 calendar years to avoid instability in During the period 1987 through 1996, the ICD-9 code, 705W, rates of rare cancers, were calculated by dividing number contains also other diseases, namely, pompholyx, bromhi- of the first primary cancers excluding those discovered in- drosis, and chromhidrosis. However, the incidence of these cidentally at autopsy by person-years at risk (number of 3 diagnoses in Swedish inpatients is very low. midyear population without reported cancer). The 95% con- fidence interval (CI) of the SIR was calculated on the as- FOLLOW-UP DATA sumption that the observed number follows a Poisson dis- tribution.18 For selected cancer sites, stratified analyses were Record linkage of the study cohort to the nationwide Reg- also performed to detect any difference of risk pattern across ister of Causes of Death, using the national registration num- sex, duration of follow-up, period at discharge, status of bers as identifiers, provided information on dates and causes comorbidities, and whether patients had ever undergone of death among those deceased through 1997. Correspond- surgical excision of the affected skin areas. An approxi- ing linkage to the emigration register identified dates of emi- mate x2 test was used to test the difference between 2 SIRs.19 gration. The National Swedish Cancer Register, founded We also calculated standardized mortality ratios for se- in 1958 and close to 98% complete,17 was used to ascer- lected causes of death. Patients with prevalent cancers tain all incident cancers until December 31, 1997. The can- (ie, those who had a previously registered diagnosis of cer register coded malignant neoplasm according to the cancer) were included in the mortality analyses. In the main ICD-7 classification during the entire study period. analyses, we excluded cancers and person-years accumu- To remove records with incorrect national registra- lated during the first year of follow-up to minimize the tion numbers, which would otherwise contribute person- possible influence of selection bias. Such bias occurs if years at no risk of cancer, we also linked the cohort file to patients with HS and a subclinical cancer are more likely the register of the total population. If a national registra- to be hospitalized than those without a subclinical cancer. tion number could not be found in this register or in the If this is the case, these cancers are most likely to be diag- death and emigration registers, we concluded that it did nosed within the first year of follow-up. This study is not correspond to an existing person. We, thus, excluded approved by ethics committee of Huddinge University from the cohort 22 medical records because of erroneous Hospital, Stockholm, Sweden. noma. Significantly elevated relative risks were also ob- No obviously increased risk was observed for colon, rec- served for buccal cancer (5 cases; SIR, 5.5; 95% CI, 1.8- tum, breast, female genital system, and brain cancers in 12.9) and primary liver cancer (3 cases; SIR, 10.0; 95% our cohort (Table 2). Nine cancers are not accounted for CI, 2.1-29.2).