Figure 1. Actuarial estimate of observed survival, comparing patients with both upper aerodigestive tract (UADT) cancer and chronic lymphocytic leukemia (CLL) and those with UADT cancer only.
Figure 2. Actuarial estimate of relative survival for patients with localized (A) and regionalized (B) upper aerodigestive tract (UADT) cancer. Patients with chronic lymphocytic leukemia (CLL) have poorer survival.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Pagedar NA, Halfdanarson TR, Karnell LH, Hoffman HT, Funk GF. Upper Aerodigestive Tract Cancer in Patients With Chronic Lymphocytic Leukemia: Incidence, Stage, and Outcome. Arch Otolaryngol Head Neck Surg. 2012;138(12):1171–1175. doi:10.1001/jamaoto.2013.723
Author Affiliations: Departments of Otolaryngology–Head and Neck Surgery (Drs Pagedar, Karnell, Hoffman, and Funk) and Medicine (Dr Halfdanarson), University of Iowa Carver College of Medicine, Iowa City.
Objective To compare incidence, stage, and survival of upper aerodigestive tract (UADT) cancers in patients with and without chronic lymphocytic leukemia (CLL).
Design Inception cohort.
Setting National database.
Patients Individuals with CLL and UADT cancers included in the Surveillance, Epidemiology, and End Results (SEER) database.
Main Outcome Measures Incidence was compared by computing standardized incidence ratio (SIR), the ratio of observed UADT cancers in patients with CLL, and the number of UADT cancers expected based on the characteristics of patients with CLL and population incidence of UADT cancers. The association between CLL and UADT cancer stage was measured using odds ratio (OR) calculations. Survival of patients with UADT cancer with and without CLL was compared.
Results For the SIR calculation, 36 985 patients with CLL contributed a mean 6.36 years of follow-up each, for a total of 235 314 person-years of follow-up. The SIR was 1.18 (95% CI, 0.97-1.41) for UADT cancers; 1.52 (95% CI, 1.18-1.93) for laryngeal cancer; and 1.92 (95% CI, 1.05-3.23) for cancers of the nasal cavity and paranasal sinuses. In the stage and survival analyses, 253 patients with CLL followed by a UADT cancer were compared with 133 840 patients with 1 UADT cancer only. Cancers of the UADT in patients with CLL were more likely localized (OR, 0.50; 95% CI, 0.37-0.68). Relative survival was worse in patients with CLL. In multivariate analysis, CLL was independently associated with poorer observed survival (hazard ratio, 1.45; 95% CI, 1.24-1.70).
Conclusions Larynx and nasal cavity cancers were more common in patients with CLL. Overall incidence of UADT cancers was not significantly elevated. Cancers of the UADT in patients with CLL were more likely to be localized at diagnosis than those in patients without CLL. Finally, CLL was associated with poorer survival outcomes.
Chronic lymphocytic leukemia (CLL) is the most common leukemia in adults in the United States, with an age-adjusted annual incidence of 4.7 per 100 000 men and 2.5 per 100 000 women. The incidence increases with age.1 Immunosuppression is commonly considered a hallmark of CLL. Between 25% and 50% of patients with CLL ultimately die of infection, most frequently involving the respiratory or urinary tracts.2 Small lymphocytic lymphoma is an indolent B-cell malignant condition, which is immunophenotypically identical to CLL but lacks the circulating lymphocytosis seen in CLL. The current World Health Organization classification considers small lymphocytic lymphoma and CLL as a single entity.3,4
Conditions causing immunosuppression are frequently associated with alterations in the behavior of neoplasms. A recent study described changes in cancer incidence in patients who had undergone a solid-organ transplant,5 and human immunodeficiency virus–related cancers have been well described. Chronic lymphocytic leukemia has been studied in this context by Travis et al,6 Hisada et al,7 and Morton et al.8 Ishibe and Curtis,9 using the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database, reported a statistically significant 19% increase in the risk of subsequent primary cancers in patients with CLL, with the lip, larynx, and nasal cavity among the primary sites demonstrating increased risk. Staging and outcomes were not reported in that study.
The purpose of this study was to describe the effect of CLL on the incidence, stage at diagnosis, and natural history of subsequent cancers of the upper aerodigestive tract (UADT). This study represents the addition of 8 years of data to the work published by Ishibe and Curtis.9 In addition, we sought to describe the stage at presentation and outcomes of UADT cancers occurring in patients with a prior diagnosis of CLL.
This study used deidentified data in a public-use data set and therefore did not require institutional review board approval. Data were obtained from SEER using SEER*Stat 7.0.5 software.10 This database includes comprehensive cancer incidence data in 9 geographic locations (Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco/Oakland, Seattle/Puget Sound, and Utah) from 1973 through 2008, in 4 more locations (Los Angeles, San Jose/Monterey, Alaska Natives, and rural Georgia) from 1992 through 2008, and in an additional 4 locations (Kentucky, New Jersey, Louisiana, and greater California) from 2000 through 2008. The 9-registry database covers approximately 10% of the US population, and the full 17-registry database covers approximately 26% of the US population.
We defined CLL and small lymphocytic lymphoma using the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) histology codes 9670 and 9823, respectively. Upper aerodigestive tract cancers were defined using ICD-O-3 primary site codes C00.0-C14.8 and C30.0-C32.9. All histologic types at UADT sites were included.
To compare incidence, SEER*Stat software was used to compute standardized incidence ratios (SIRs) for UADT sites from population incidence rates adjusted for age, race, and sex. This inquiry was limited by SEER*Stat to the 9-registry database. The SIR compares the number of observed cases to the number of cases expected based on the population characteristics of patients with CLL, the incidence of UADT cancer in specified populations, and the number of years of follow-up per patient with CLL. The standard SEER rate files were modified to classify the tongue base as an oropharynx site, in contrast to its ICD-O-3 topographic classification in the oral cavity. To minimize the possibility of including UADT cancers that existed at the time the CLL was diagnosed, UADT cancers were included only if diagnosed 2 or more months after the CLL.
The remainder of the analysis was performed on data from the full 17-registry SEER database over the years 1973 through 2008, using SAS 9.2 software (SAS Institute Inc). Stage at presentation, treatment, and survival outcomes were compared between 2 cohorts: patients who had CLL and a subsequent UADT cancer and those who had a single UADT cancer and no other cancers. Odds ratios (ORs) for the association between CLL and SEER stage (local/regional/distant) were calculated, with Mantel-Haenszel adjustment for primary site. The χ2 test was used to evaluate statistical significance. Odds ratios were calculated using patients with known stage.
Multivariate Cox proportional hazards analysis was used to investigate the effect of CLL on observed survival, correcting for age, stage, primary site, and year of diagnosis. We sought to correct for the effect of deaths unrelated to cancer by comparing relative survival based on stage. The reliability of cause-of-death data in registries, including SEER, is questionable, especially for UADT cancers.11 We therefore used relative survival, rather than cause-specific survival, to evaluate the survival effect of CLL in patients with UADT cancer. Relative survival compares survival observed in the cohort of interest with that expected in an otherwise identical cohort without cancer.12 Expected survival is computed from US population data, accounting for the age, year, sex, and race composition of the cohort of interest. While calculation of cause-specific survival depends on the accuracy of individual death certificates, the SEER*Stat calculation of relative survival relies on the assumption that the population mortality of the cancer of interest is small, so that the survival outcomes of the entire US population are used to approximate those of a cohort without cancer. SEER*Prep 2.4.5 software (http://seer.cancer.gov/seerprep/) was used to create an analytic data set from the 2 cohorts (one with UADT cancer alone and the other with both UADT cancer and CLL). For each stage, SEER*Stat was then used to compute relative survival. Statistical comparison of relative survival estimates was performed using a semiparametric model with CANSURV 1.0 (Statistical Research and Applications Branch, National Cancer Institute) as described by Gamel et al.13
During the study period, 36 985 patients were diagnosed as having CLL, with 235 314 person-years of follow-up, representing a mean of 6.36 follow-up years per patient. The resulting SIR data are given in Table 1. The SIR for cancers of all UADT sites was 1.18 (95% CI, 0.97-1.41). In other words, the incidence of UADT cancer was 18% greater in patients with CLL. The SIR for laryngeal cancer was 1.52 (95% CI, 1.18-1.93), and for nasal cavity and paranasal sinus sites, 1.92 (95% CI, 1.05-3.23).
To account for potential changes in incidence of oropharynx cancer related to human papillomavirus, separate SIRs were calculated for the periods 1973 through 1995 and 1996 through 2008. The SIR for oropharynx cancer for the period 1973 through 1995 was 0.91 (95% CI, 0.51-1.49), and in the later period, 0.44 (95% CI, 0.16-0.95). The latter was statistically significant, although with 13.7 oropharynx cancers expected, the sample size was small.
The full 17-registry SEER database contained 253 patients with CLL and a subsequent head and neck cancer. There were 133 840 patients with 1 head and neck cancer and no other cancers. Comparisons of the 2 cohorts are given in Table 2. Patients with CLL were more likely to be male and were on average a decade older at the time of diagnosis of UADT cancer. As given in Table 2, the overall distribution of primary sites within the UADT differed between the groups, with the most clinically significant differences being the more frequent laryngeal cancers and less frequent oropharynx cancers in the CLL group. Importantly, these data were in agreement with the SIR data reported in Table 1, which were derived from the 9-registry database.
Also given in Table 2 is the distribution of UADT cancer stage at presentation. Head and neck cancers were less likely to demonstrate regional or distant spread in patients with CLL, with an OR of 0.40 (95% CI, 0.30-0.54) (P < .001). Primary site may well be a confounder in the association between CLL and stage at presentation, so the Mantel-Haenszel OR was calculated. After adjusting for primary site, the OR was 0.50 (95% CI, 0.37-0.68) (P < .001) (Breslow-Day test of homogeneity, P = .58, indicating validity of the adjusted OR).
We used follow-up data to compute actuarial observed survival estimates comparing patients with CLL and those without CLL. The estimates are shown in Figure 1. The survival estimates for the CLL and the non-CLL group were different (log-rank, P < .001). A multivariate Cox proportional hazard model was developed to evaluate the effect of CLL on observed survival. The model also included age, stage, primary site, and year of diagnosis. The results are given in Table 3. The hazard ratio for CLL was 1.45 (95% CI, 1.24-1.70) (P < .001). The proportional hazard assumption was verified (data not shown).
We generated separate relative survival comparisons data for local, regional, and distant cancers. For each stage, the difference in relative survival between the CLL and non-CLL groups was significant (P < .001). The relative survival estimates for local and regional UADT cancers are shown in Figure 2.
These data represent an update of earlier studies with additional data on stage and survival. The inclusion of data from 2000 through 2008 adds significantly to the report of Ishibe and Curtis.9 Specifically, the latter study included more than 19 000 patients followed for approximately 115 000 person-years. The present report more than doubles the sample size. Two other national registry-based studies have reported standardized incidence ratios for second primary UADT cancers in patients with a history of CLL. Schöllkopf et al,14 studying the Danish cancer registry, did not identify altered incidence of UADT cancers. Royle and colleagues15 analyzed second cancers in patients with CLL using an Australian cancer registry and reported an increase in the risk of cancers of the oral cavity, pharynx, and larynx. The present study represents includes substantially more patients and longer follow-up. It is not clear why the 3 registries would produce different findings.
The reasons for the effect of CLL on UADT cancer incidence are not known. Immunosuppression associated with CLL, either intrinsic to the CLL itself or related to the treatments used, is a primary consideration. The UADT sites, which are more susceptible, are not ones typically associated with virally mediated carcinogenesis. In fact, given the incidence findings in the later-period (1996-2008) cohort, there may be a protective effect in the oropharynx, where human papillomavirus is known to be a causative factor. Hisada et al7 proposed a role for shared risk factors, such as tobacco use, since cancers of the lung occurred at a higher-than-expected rate. However, tobacco use is also a strong risk factor for oral cancer, which was not more likely in patients with CLL.
Our study identified differences in stage at presentation, with UADT cancers more likely to be localized in patients with CLL. A possible explanation for this is a surveillance bias. Typically, patients with CLL maintain regular follow-up with a medical oncologist 2 to 4 times per year and therefore might be more likely to receive referral for evaluation of UADT-related symptoms that might permit earlier diagnosis.
The SEER data set did not permit us to compare how UADT cancers were treated in patients with CLL and in those without CLL. The SEER database provides reliably complete information about surgical treatment and radiotherapy. However, matching chemotherapy information is not available. Without accounting for the use of chemotherapy, it is not possible to usefully compare treatment regimens for UADT cancers or assess the effect of therapy directed at CLL.
The relationship between CLL and squamous and basal cell carcinomas of the skin is of interest to head and neck oncologists. That relationship is not described in this study because those common cutaneous malignant conditions are not reported to SEER.
A significant issue facing this group of patients is perioperative morbidity and mortality. Our study does not address this topic. Immunosuppression is frequently considered a powerful risk factor for postoperative wound-healing complications. Unfortunately, the SEER database does not contain variables that would permit assessment of this relevant question. Additional studies involving review of patient medical records would be required to address this important issue.
In conclusion, the experience of UADT carcinoma is different in patients with a history of CLL. Incidence rates are higher than expected in the larynx and in the nasal cavity and paranasal sinuses. Staging data demonstrate a likely surveillance effect, with patients with CLL presenting at earlier stage. Chronic lymphocytic leukemia is an independent risk factor for mortality in patients with UADT cancer.
Correspondence: Nitin A. Pagedar, MD, Department of Otolaryngology–Head and Neck Surgery, University of Iowa, 200 Hawkins Dr, Iowa City, IA 52242 (firstname.lastname@example.org).
Submitted for Publication: July 12, 2012; final revision received August 21, 2012; accepted September 7, 2012.
Author Contributions: Dr Pagedar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Pagedar, Halfdanarson, and Hoffman. Acquisition of data: Pagedar and Karnell. Analysis and interpretation of data: Pagedar, Halfdanarson, Karnell, Hoffman, and Funk. Drafting of the manuscript: Pagedar, Halfdanarson, Hoffman, and Funk. Critical revision of the manuscript for important intellectual content: Halfdanarson, Karnell, Hoffman, and Funk. Statistical analysis: Pagedar, Halfdanarson, and Karnell. Study supervision: Hoffman and Funk.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented as a poster at the Eighth International Head and Neck Cancer Conference; July 22, 2012; Toronto, Ontario, Canada.
Create a personal account or sign in to: