RESULTS
Autopsy reports classically document a “Final Diagnosis” that describes cancer primary site, histologic findings, and tumor expansion and metastases based upon an examination of all major organs.9 If an autopsy report first documents the presence of cancer post-mortem, documentation is reviewed to ensure that no clinical cancer diagnosis was made prior. If no diagnosis is listed, the case is identified “diagnosed at autopsy.”9Autopsy reports outline a succession of events that convey essential information in order to provide the useful information regarding the patient’s condition including but not limited to chain of events, injuries, or complications. 10 In cases where cause of death is not apparent despite a thorough autopsy, a judgment may be made by the coroner.10 The functional definition of “cause of death” is important to understand in the context of autopsy reporting because cancer can be an incidental finding that could lead to misreporting of cancer-related deaths.
Sesterhenn et al. reviewed the cases of 91 HNC-related deaths from 1968 to 2007 and showed distant metastases in 46.2% and second primary tumors in 17.6% of the cases reviewed.11 Multiple studies have revealed substantial rates of discrepancy between pre- and post-mortem diagnoses often varying from 10% up to 40%.12 Another case report discusses two patients who presented clinically with no evidence of disease and were considered ‘cured,’ but had significant distant spread of disease upon autopsy.13 Misdiagnoses that may have altered patient treatment course have been quoted to range anywhere from 2.4% to 10.7%.12
Of non-thyroid HNCs, laryngeal cancer is an uncommon subtype, and its prevalence is decreasing over time.14 Despite this, it was the most common non-thyroid HNC diagnosed upon autopsy in our study and was the primary cause of death in half of these patients. It is unlikely these patients were asymptomatic as 50% of tumors were stage 3 and 4; thus, some combination of hoarseness, odynophagia, otalgia, neck lumps, or mechanical disruptions in swallowing or breathing would likely be present. It is unclear why the prevalence of undiagnosed laryngeal cancer was high in our study compared to other HNCs. It is unlikely socioeconomic status is a major contributing factor as most of the patients examined fell into the same middle-class stratum of annual income. Rather than misdiagnosis or socioeconomic factors, it is possible patients who participate in heavy tobacco and alcohol use, well-known risk factors for laryngeal cancer, are less likely to seek medical care for any number of biopsychosocial reasons. Smoking is associated with reduced social support, decreased intrinsic view of self-efficacy, low health-related motivation, and a decreased likelihood to complete pharmacologic or psychological treatment regimens.15 Considering 27.77% of the patients in our study were over 70 years old, it is also plausible that primary care physicians could interpret symptoms of non-thyroid HNC as part of the natural aging process due to an unfamiliarity of these pathologies and their clinical presentations.
In sum, 47.06% of patients with non-thyroid HNC-related deaths were found in patients with stage 1 or 2 cancers, which raises concerns regarding documentation. Currently, otolaryngologists do not consider stage 1 and 2 HNCs to be highly lethal. These results appear to contradict that sentiment. Although autopsy is a powerful tool in cancer biology, delays in autopsy and refrigeration of body tissue can decrease the integrity of the tumor due to tissue decomposition. In addition to macroscopic changes to tumors, delays in autopsy allow for molecular alteration which may change the morphology of cancer cells.16 Though small, these changes could affect tumor grading and lead to underreporting of disease severity. Additionally, because of the underuse of post-mortem imaging techniques if a full-body autopsy is not completed, it is conceivable that not all metastatic sites of disease will be discovered and reported; as of 2008, an autopsy was performed in under 10% of all deaths in the United States.17
Scarce autopsies could further the possibility of underreporting undiagnosed HNC. It is possible that stage 1 and 2 cancers have a higher-than-expected death rate because autopsies are usually conducted in instances of unknown causes of death and pathologists may assume incidental findings as causes of deaths if no other obvious pathologies are found. All these factors can contribute to underdiagnosis or misdiagnosis of cancer severity, which could cause lower stages of non-thyroid HNCs to appear more malignant.
Non-thyroid HNC is over 17 times more likely to be designated as the primary cause of death compared to thyroid cancer in patients with undiagnosed HNCs at autopsy. One plausible explanation for this finding is that 50.00% of laryngeal cancers found at autopsy were stage 3 or 4 while only 11.81% thyroid carcinoma were stage 3 or 4. Patients with thyroid cancer had relatively similar death rates in primary cause of death categories as the general public.18 This data suggests thyroid cancer did not contribute to the mortality of these patients and was an incidental finding that may or may not had associated symptoms. It should be noted that not all thyroid cancers are latent. Medullary, anaplastic, and Hurthle cell thyroid cancers are often more aggressive in nature; however, these cancers only contributed to 3.18% of the cancers analyzed. We also found that patients over 70 years old were at lower risk of thyroid related deaths than those under 50, likely because other health issues such as cardiovascular disease are more common causes of deaths in elderly populations. Conversely, non-thyroid cancer is often more aggressive with poorer prognoses.19Consequentially, non-thyroid cancers are more likely to be the cause of death than thyroid cancer.
Fewer patients with non-thyroid HNC were found at autopsy compared to thyroid cancer, likely due to the higher prevalence of thyroid cancer and decreasing incidence of HNC overall.20,21Our study found no identifiable temporal trends in detection of non-thyroid cancers upon autopsy; however, there was a discernible decrease in thyroid cancers diagnosed at autopsy over time. This is likely due to increased screening and biopsy of thyroid cancer in recent decades.10 This study demonstrates the malignancy of non-thyroid cancers and their prevalence as the primary cause of death upon autopsy in patients previously undiagnosed. These findings demonstrate the need for the medical community’s awareness of their prevalence in asymptomatic patients and the need for screening in those high risk for disease development or progression.
There are inherent limitations within the dataset obtained from SEER that includes miscoding and omitting data especially in variables such as TMN staging and tumor size. This dataset has a limited number of reviewable patients and autopsy reports that may or may not be representative of the total population of the United States. Because the SEER database only covers about 28% of the U.S. population, the results may not be completely generalizable.22 In addition, many of the variables within SEER have a significant number of data points missing; therefore, non-statistically significant findings may not be truly accurate.