2.1

Cohort definition

National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) Program was used for obtaining data. The SEER Program is the principal source for cancer statistics and represents 28% of the US population. The currently operating 18 SEER registries were utilized to procure information on demographics, incidence and survival. SEER data was collected and reported using data items and codes as documented by the North American Association of Central Cancer Registries (NAACCR) . All neoplasms reported to the SEER Program since 2001 have been coded using the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) and those diagnosed prior to 2001 have been recoded to this classification.

Invasive malignancies classified as adenoid cystic carcinomas (ICD-O-3 codes 8200/3) originating from BOT were included. Cases were excluded from the analyses if they were identified from death certificates only, were of unknown age, or were not confirmed histologically. All patients with ACC of the BOT reported during the study period from 1973 to 2012 were characterized according to age, gender, race, marital status, grade, historical stage A, year of diagnosis and tumor-directed treatment. Marital status was classified into married, single (which included never married, separated, divorced and widowed) and unknown. Grade was classified into well differentiated, moderately differentiated, poorly differentiated and undifferentiated. Tumor-directed therapy was grouped into 4 categories: surgery alone, radiation alone, combined modality (surgery and radiation) and unknown treatment.

Cancer staging has changed tremendously from 1970s to the present years; TNM has only been available in the SEER database since 2004. Hence, we used the “SEER historic stage A” for staging, since it covered the entire study period. It groups cancer cases as localized, regional, distant, or unstaged based on the following definition

• •
  

  Stage of localized cancer – Cancer that is limited to the organ of origin with no evidence of spread beyond it.

  
  
• •
  

  Stage of regional cancer – Cancer that has spread beyond the organ of origin (primary site) to surrounding lymph nodes, organs or tissues with no spread to distant organs.

  
  
• •
  

  Stage of distant cancer – Cancer that has spread beyond organ of origin to distant lymph nodes, organs or tissues remote from the organ of origin.

  
  
• •
  

  Unstaged cancer – Not enough information to indicate a stage.

2.2

Statistical analysis

Incidence rates by sex and race were calculated for the entire study period, and expressed as the number of cases per 1,000,000 individuals per year, along with 95% confidence intervals (CIs). Rate ratios (RRs) along with 95% CIs were utilized to compute the gender and race differences . The survival of patients with ACC was analyzed using several methods, and the total follow-up was extended to 20 years post diagnosis. A 5, 10, 15, and 20-year cumulative observed survival (OS) defined as the proportion of patients who survived beyond a given interval was calculated. Then the relative survival (RS) obtained by dividing OS among cancer patients by the expected survival in the general population with the same age, race, and sex characteristics was computed. The survival analyses excluded cases that represented second or later primary cancer, unlike frequency analyses, which included all records. Patients known to be alive but with no documentation of their survival time were excluded (resulting in a total population of n = 187). The 5, 10, 15, and 20-year OS and RS estimates were evaluated overall, as well as by gender, race and stage. The follow-up data for survival analyses extended through to the end of 2012.

Survival analysis according to historical stage A and tumor-directed treatment was determined over a 10-year interval, while the follow-up for survival analyses by decade was limited to 5-years to allow comparisons between earlier and later intervals. We constructed Kaplan-Meier curves with corresponding log-rank (Mantel-Cox) tests in order to determine the statistical significance of patient survival according to historical stage A and type of therapy. Additionally, we used multivariable Cox proportional hazard models to examine the association between survival and various patient-related, disease-related and treatment-related characteristics with the primary focus on disease stage and treatment type. The results of these multivariable models are expressed as adjusted hazard ratios (HRs) and are reported along with the corresponding 95% CIs and p -values. All p -values were determined to be significant if < 0.05. All data was analyzed using SAS version 9.3 (SAS institute Inc. Cary, NC) and SEER*Stat version 8.2.1 (National Cancer Institute, Bethesda, Md) statistical software packages.

2.1

Cohort definition

National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) Program was used for obtaining data. The SEER Program is the principal source for cancer statistics and represents 28% of the US population. The currently operating 18 SEER registries were utilized to procure information on demographics, incidence and survival. SEER data was collected and reported using data items and codes as documented by the North American Association of Central Cancer Registries (NAACCR) . All neoplasms reported to the SEER Program since 2001 have been coded using the International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) and those diagnosed prior to 2001 have been recoded to this classification.

Invasive malignancies classified as adenoid cystic carcinomas (ICD-O-3 codes 8200/3) originating from BOT were included. Cases were excluded from the analyses if they were identified from death certificates only, were of unknown age, or were not confirmed histologically. All patients with ACC of the BOT reported during the study period from 1973 to 2012 were characterized according to age, gender, race, marital status, grade, historical stage A, year of diagnosis and tumor-directed treatment. Marital status was classified into married, single (which included never married, separated, divorced and widowed) and unknown. Grade was classified into well differentiated, moderately differentiated, poorly differentiated and undifferentiated. Tumor-directed therapy was grouped into 4 categories: surgery alone, radiation alone, combined modality (surgery and radiation) and unknown treatment.

Cancer staging has changed tremendously from 1970s to the present years; TNM has only been available in the SEER database since 2004. Hence, we used the “SEER historic stage A” for staging, since it covered the entire study period. It groups cancer cases as localized, regional, distant, or unstaged based on the following definition

• •
  

  Stage of localized cancer – Cancer that is limited to the organ of origin with no evidence of spread beyond it.

  
  
• •
  

  Stage of regional cancer – Cancer that has spread beyond the organ of origin (primary site) to surrounding lymph nodes, organs or tissues with no spread to distant organs.

  
  
• •
  

  Stage of distant cancer – Cancer that has spread beyond organ of origin to distant lymph nodes, organs or tissues remote from the organ of origin.

  
  
• •
  

  Unstaged cancer – Not enough information to indicate a stage.

2.2

Statistical analysis

Incidence rates by sex and race were calculated for the entire study period, and expressed as the number of cases per 1,000,000 individuals per year, along with 95% confidence intervals (CIs). Rate ratios (RRs) along with 95% CIs were utilized to compute the gender and race differences . The survival of patients with ACC was analyzed using several methods, and the total follow-up was extended to 20 years post diagnosis. A 5, 10, 15, and 20-year cumulative observed survival (OS) defined as the proportion of patients who survived beyond a given interval was calculated. Then the relative survival (RS) obtained by dividing OS among cancer patients by the expected survival in the general population with the same age, race, and sex characteristics was computed. The survival analyses excluded cases that represented second or later primary cancer, unlike frequency analyses, which included all records. Patients known to be alive but with no documentation of their survival time were excluded (resulting in a total population of n = 187). The 5, 10, 15, and 20-year OS and RS estimates were evaluated overall, as well as by gender, race and stage. The follow-up data for survival analyses extended through to the end of 2012.

Survival analysis according to historical stage A and tumor-directed treatment was determined over a 10-year interval, while the follow-up for survival analyses by decade was limited to 5-years to allow comparisons between earlier and later intervals. We constructed Kaplan-Meier curves with corresponding log-rank (Mantel-Cox) tests in order to determine the statistical significance of patient survival according to historical stage A and type of therapy. Additionally, we used multivariable Cox proportional hazard models to examine the association between survival and various patient-related, disease-related and treatment-related characteristics with the primary focus on disease stage and treatment type. The results of these multivariable models are expressed as adjusted hazard ratios (HRs) and are reported along with the corresponding 95% CIs and p -values. All p -values were determined to be significant if < 0.05. All data was analyzed using SAS version 9.3 (SAS institute Inc. Cary, NC) and SEER*Stat version 8.2.1 (National Cancer Institute, Bethesda, Md) statistical software packages.