Underreporting of Vestibular Schwannoma Incidence Within National Brain Tumor and Cancer Registries in the United States

Abstract

Objective:

Reports using large brain tumor and cancer registries suggest that the incidence of vestibular schwannoma is considerably lower in the United States compared with other countries. The current study compares the incidence and disease characteristics of vestibular schwannoma within a large tertiary referral center’s brain tumor and cancer registry using nationally mandated reporting protocols to disease incidence in the same population using an externally validated population-based consortium.

Study Design:

Population-based study spanning 1,945,007 person-years.

Setting:

Large tertiary referral center.

Patients:

Adults with sporadic vestibular schwannoma.

Main Outcome Measure:

Disease incidence rates from 2004 to 2016.

Results:

From 2004 to 2016, the incidence of vestibular schwannoma in the tumor registry was 1.3 per 100,000 person-years whereas the population-based cohort had an incidence of 4.4 per 100,000 person-years. From 2012 to 2016, the incidence in the tumor registry was 1.4 per 100,000 person-years compared with 5.2 in the population-based cohort. Patients within the population-based cohort were significantly more likely to have smaller tumors at diagnosis (78% intracanalicular versus 45%; p = 0.004) and consequently more likely to undergo management consisting of observation with serial imaging as opposed to treatment with either microsurgery or radiosurgery (71% versus 28%; p = 0.001).

Conclusions:

The reliance on pathology specimens and cancer-related treatment data for the national registration of new cancer and brain tumor diagnoses may introduce selection bias and underreporting of benign brain tumors that frequently involve observation as a primary treatment modality. This selection bias likely accounts for the discrepant incidence rates of vestibular schwannoma reported between the United States and other countries.

Countries with national healthcare systems readily allow for centralized patient registries. However, the United States’ decentralized healthcare system poses significant challenges to conducting robust epidemiologic research. To address this, large population-based consortiums, such as the National Cancer Database (NCDB) or the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute, have been established (1). In 2004, the Benign Brain Tumor Cancer Registries Amendment Act (public law 107–260) mandated the national registration of benign brain tumors. As a result, healthcare institutions are now required to have registry infrastructure that systematically identifies and reports patients diagnosed with benign brain tumors in addition to new cancer diagnoses. Given the vast number of different types of benign brain tumors and malignancies, patient registration through exhaustive manual searches of clinic visits is logistically impractical for most institutions. Therefore, institutions often have infrastructure that queries pathology reports (e.g., biopsied tissue, ablative surgical tissue) and cancer-related treatment codes, such as those related to radiation and chemotherapy, to identify most new diagnoses.

Although most institutions employ the same identification and reporting mechanisms for both benign brain tumors and cancer diagnoses, the clinical pathways with which cancer patients are diagnosed and treated fundamentally differ from those associated with benign brain tumors. For instance, the diagnosis of cancer requires obtaining a tissue specimen, and cancer patients typically receive multidisciplinary therapeutic management that involves surgery, radiation, chemotherapy, or a combination thereof. By contrast, most benign brain tumors are diagnosed clinically with neuroimaging and rarely undergo biopsy. Moreover, patients often undergo a period of observation with serial imaging following diagnosis and may never require definitive microsurgical or radiosurgical treatment (2). For many institutions, the registration of patients who only have a “clinical diagnosis” of their disease (i.e., no pathology specimen)—as is commonplace for many benign brain tumors—is more challenging. In these ways, since most institutions’ tumor registration infrastructure is designed to identify the wide array of cancer diagnoses, the unique characteristics of benign brain tumors may render their registration less effective.

Incidence rates for vestibular schwannoma in the United States contradict global incidence rates. Numerous reports from SEER and the Central Brain Tumor Registry of the United States (CBTRUS) have suggested that the national incidence of vestibular schwannoma is around 1.0 to 1.9 per 100,000 person-years and has been largely stable since 2004 (3–5). However, as early as the mid-2000s, regions of the United States with elevated access to healthcare (e.g., Beverly Hills, CA) reported incidence rates exceeding five per 100,000 person-years (6). More recently, a population-based study reported that the incidence of vestibular schwannoma was 4.2 per 100,000 person-years from 2006 to 2016 (7). For context, these estimates exceed recently published incidence rates of pituitary adenoma (5). National data from Denmark corroborate these findings, reporting an incidence of vestibular schwannoma of 3.4 per 100,000 person-years in 2015 with significant increases over the previous decade and an over 10-fold increase since 1976 (8). Similarly, a recent publication covering the Taiwanese population reported a peak incidence in recent years of 3.7 per 100,000 person-years (9). In addition to discrepant incidence rates, many of these population-based studies report patient demographic and clinical features that differ from data published from large national cancer registries in the United States. For instance, reported data from the NCDB, SEER, and CBTRUS, describe patient cohorts that often have a younger median patient age at time of diagnosis, larger tumor size at time of diagnosis, and a higher proportion of patients who pursue definitive treatment with either microsurgery or radiosurgery (3,4,8).

Direct study into the potential reasons for lower reported incidence rates of vestibular schwannoma within large national registries is difficult as there exists no “gold standard” for comparison. Instead, to understand the discrepancies between published data, a comparison of cases reported from nationally mandated protocols and an externally validated population-based database that both identify patients from the same geographic region may illuminate differences in registration and reporting. The current study compares the incidence of vestibular schwannoma obtained from institutionally identified cases from nationally mandated protocols established by the Benign Brain Tumor Cancer Registries Amendment Act of 2004 to the incidence in the same population obtained from manual review of all clinical visits associated with a diagnosis of vestibular schwannoma using an externally validated population-based medical records-linkage system (10,11).

MATERIALS AND METHODS

Clinical Data

After obtaining institutional review board approval (IRBs 15–006036, 16–007363, and 050-OMC-15), all patients diagnosed with vestibular schwannoma from January 1, 2004 through December 31, 2016 from both the Mayo Clinic Cancer Registry and the Rochester Epidemiology Project database were reviewed.

Since 2004, the Mayo Clinic Cancer Registry has systematically identified and registered patients diagnosed with benign brain tumors in addition to cancers. As is standard practice, vestibular schwannoma was coded using the International Classification of Diseases (ICD) for Oncology codes for schwannomas (9560/0) with collaborative staging coding for the acoustic nerve (72.4) in accordance with the Minnesota Department of Health Cancer Reporting stipulations. Most relevant coding rules are governed by the North American Association of Cancer Registries. At our institution, Health Level 7 listener software linked to institutional pathology reports searches approximately 3,500 terms and abbreviations that could refer to a benign brain tumor or cancer diagnosis. These cases are marked for further review by cancer registry personnel. Although patients diagnosed with a brain tumor or cancer can also be identified through radiation and chemotherapy-related procedural codes, as well as relevant ICD-9 or ICD-10 diagnostic codes, manual searches of all relevant diagnoses from clinical encounters are not routinely performed due to the overwhelming number of search returns and lack of specificity of these search results. Registry data abstraction for the current study was performed independently by Mayo Clinic Cancer Registry personnel using the aforementioned codes with additional filtering to identify patients who received their first lifetime diagnosis of vestibular schwannoma while a resident of Olmsted County, Minnesota. Each patient’s medical record was subsequently reviewed for patient-level data. Analogous application of the “intention-to-treat” analysis was employed in an effort to most closely approximate data abstraction limitations of large population-based brain tumor and cancer registries in the United States. As a byproduct, patients who were found to have non-sporadic vestibular schwannoma (i.e., neurofibromatosis type 2 [NF2]) or were incorrectly coded as having vestibular schwannoma (e.g., other cranial nerve schwannomas) were included in the analysis.

Supported by the National Institute on Aging, the Rochester Epidemiology Project is a unique medical records-linkage system within the United States that covers a complete population in a well-defined geographic region since 1966 (10–12). Several previous studies have validated the reputability of the Rochester Epidemiology Project (12,13). There exists consistent clinical follow-up among the population with almost 90% of residents age 50 years seen by a healthcare provider yearly, and this number approaches almost 100% for those over age 70 years (13). Cases of vestibular schwannoma were identified after a diagnosis by a medical provider during a healthcare encounter within the geographic region or were found through search of relevant procedure codes, pathology reports, or autopsy findings. Specific codes for patients identified in the current study included ICD-9 code 225.1 and ICD-10 code D33.3. In addition, Mayo Clinic had its own coding system for research purposes that was in place from 1976 to 2010, and this database is searched by the Rochester Epidemiology Project. A total of 222 cases from January 1, 2004 through December 31, 2016 were initially retrieved, and a manual review of the patient’s complete medical record from almost every healthcare encounter within the geographic region was performed to identify incident cases of sporadic vestibular schwannoma. An analysis of the epidemiologic trends of vestibular schwannoma in this population has been previously reported (7,14).

Statistical Methods

Comparisons of features between cohorts were evaluated using Wilcoxon rank sum, χ², and Fisher exact tests. Tumor size was reported in accordance with the American Academy of Otolaryngology–Head and Neck Surgery committee guidelines (15). Incidence rates per 100,000 person-years were calculated using incident cases of vestibular schwannoma as the numerator and age- and sex-specific counts of the population of Olmsted County, Minnesota obtained from a complete enumeration of the Olmsted County population by the Rochester Epidemiology Project as the denominator. Incidence rates were directly standardized to the total United States population from the 2000 United States Census. Statistical analyses were performed using version 9.4 of the SAS software package (SAS Institute; Cary, NC). p-Values <0.005 were considered statistically significant.

RESULTS

Clinical Characteristics

From 2004 to 2016, the institutional tumor registry identified 25 incident cases of vestibular schwannoma within Olmsted County using abstraction methods that parallel the NCDB, SEER, and CBTRUS registries, while population-based data from the Rochester Epidemiology Project covering the same population identified 86 cases. Patients identified by the Rochester Epidemiology Project infrastructure displayed similar demographic features (Table 1), but had significantly smaller tumors at diagnosis, as evidenced by the bias towards tumors confined entirely to the internal auditory canal at time of diagnosis in this cohort (78% intracanalicular versus 45%, p = 0.004; Table 1 Supplemental Digital Content, http://links.lww.com/MAO/B193). Even amongst intracanalicular tumors, the population-based cohort median tumor size was smaller (0.8 mm versus 0.4 mm, p = 0.005). Consequently, patients identified using the Rochester Epidemiology Project were significantly more likely to undergo management consisting of observation with serial imaging (i.e., watch-and-wait) opposed to treatment with either microsurgery or radiosurgery (71% versus 28%, p = 0.001).

TABLE 1.

Clinical characteristics of patients diagnosed with vestibular schwannoma using the institutional brain tumor and cancer registry data and the Rochester Epidemiology Project^a

	REP N = 86	Subgroup of REP not in Cancer Registry N = 65	Cancer Registry N = 25
Feature		Median (IQR)		p-Valueb
Age at diagnosis (yrs)	62 (52–71)	62 (52–71)	58 (51–66)	0.28
Sex		n (%)
Female	45 (52)	34 (52)	15 (60)	0.51
Male	41 (48)	31 (48)	10 (40)
Race
White	79 (92)	59 (91)	24 (96)	0.67
Non-white	7 (8)	6 (9)	1 (4)
Tumor laterality
Left	41 (48)	31 (48)	12 (48)	0.98
Right	45 (52)	34 (52)	13 (52)
Initial treatment
Observation	48 (56)	46 (71)	7 (28)	0.001
Microsurgery	18 (21)	11 (17)	10 (40)
Radiosurgery	20 (23)	8 (12)	8 (32)

^aNo cases of sporadic vestibular schwannoma were identified by the institutional brain tumor and cancer registry that were not registered within the REP. However, there were four patients within the registry who were not included in the REP due to either not having sporadic vestibular schwannoma (i.e., NF2 or other cranial nerve schwannomas; n = 3) or their true diagnosis date was before 2004 (n = 1).

^bp-Values reflect comparisons of the subgroup of cases in the REP, but not in the brain tumor and cancer registry, and cases identified by the registry.

IQR indicates interquartile range; REP, Rochester Epidemiology Project.

Incidence Data

Using an “intention-to-treat” analysis, the incidence of vestibular schwannoma from 2004 to 2016 among cases identified in the brain tumor and cancer registry was 1.4 per 100,000 person-years, whereas the incidence among the Rochester Epidemiology Project cohort was 4.4 (Table 2). Over the last decade, the incidence among cases identified in the brain tumor and cancer registry stayed largely stable at 1.2 per 100,000 person-years from 2004 to 2007 to 1.4 from 2012 to 2016. By contrast, the incidence among the population-based cohort increased from 3.3 per 100,000 person-years to 5.2 over the same period. Excluding two cases of non-sporadic vestibular schwannoma (i.e., NF2-related) and one case of jugular foramen schwannoma, the incidence of sporadic vestibular schwannoma in the brain tumor and cancer registry dropped to 1.2 per 100,000 person-years from 2012 to 2016 (Table 2 Supplemental Digital Content, http://links.lww.com/MAO/B193).

TABLE 2.

Incidence rates of vestibular schwannoma determined by the institutional brain tumor and cancer registry and the Rochester Epidemiology Project

	Women			Men			Total
Year	N	Population	Incidencea	N	Population	Incidencea	N	Population	Incidencea
REP
2004–2007	10	297,429	3.4 (3.3b)	9	273,568	3.3 (3.5b)	19	570,997	3.3 (3.4c)
2008–2011	13	312,898	4.2 (4.0b)	14	285,649	4.9 (5.0b)	27	598,547	4.5 (4.4c)
2012–2016	22	405,641	5.4 (4.7b)	18	369,822	4.9 (4.6b)	40	775,463	5.2 (4.6c)
2004–2016	45	1,015,968	4.4(4.1b)	41	929,039	4.4 (4.4b)	86	1,945,007	4.4 (4.2c)
Cancer Registry
2004–2007	5	297,429	1.7 (1.6b)	2	273,568	0.7 (0.7b)	7	570,997	1.2(1.2c)
2008–2011	4	312,898	1.3 (1.1b)	3	285,649	1.1 (1.0b)	7	598,547	1.2(l.lc)
2012–2016	6	405,641	1.5 (1.3b)	5	369,822	1.4 (1.2b)	11	775,463	1.4(1.2c)
2004–2016	15	1,015,968	1.5 (1.3b)	10	929,039	1.1 (1.0b)	25	1,945,007	1.3(1.2c)

^aIncidence per 100,000 person-years.

^bIncidence per 100,000 person-years directly standardized by age to the 2000 US Census total population.

^cIncidence per 100,000 person-years directly standardized by age and sex to the 2000 US Census total population.

REP indicates Rochester Epidemiology Project.

DISCUSSION

With interspersed reports of incidence rates that exceed five per 100,000 person-years and robust international data showing divergent epidemiologic trends, data abstracted from large population-based brain tumor and cancer registries in the United States have appeared to consistently underreport the incidence of vestibular schwannoma (Table 3). The current work aimed to compare the incidence of vestibular schwannoma using institutionally identified cases from nationally mandated protocols to the incidence in the same population obtained following manual review of all clinical visits associated with a diagnosis of vestibular schwannoma using an externally validated population-based database. Institutional brain tumor and cancer registry incidence rates closely paralleled large brain tumor and cancer registries in the United States (4), and these data underreported disease incidence in the same population by almost fourfold. Similar to national cancer registry data, the institutional cancer registry data demonstrated a propensity for patients to present with larger tumors that required microsurgical or radiosurgical management when compared with other population-based studies (8,16–18).

TABLE 3.

Trends in global incidence rates of vestibular schwannoma published in the English literature since the early 2000s

	Incidence Circa 2004	Most Recent Incidencea
United States
SEER (4)	1.4	1.3
CBTRUS (3,5)	1.0	1.9
REP (7)(25)	3.3	5.2
Europe
Denmark (8)	2.3	3.4
Leiden region of Netherlands (19)	2.6	3.3
Asia
Taiwan (9)	1.7	3.0

^aIncidence per 100,000 person-years. Most recent year varies by publication, ranging from 2012 for the Netherlands and Taiwan, 2015 for SEER and Denmark, and 2016 for the REP and CBTRUS.

CBTRUS indicates Central Brain Tumor Registry of the United States; REP, Rochester Epidemiology Project; SEER, Surveillance, Epidemiology, and End Results.

It is likely that the observed selection bias and discrepant incidence rates stem from the methodology with which patients are registered in each database. The institutional registration system relies heavily on automated searches of diagnoses from pathology specimens due to the immense volume of patient medical record data that must be searched to identify every type of brain tumor and cancer. Although this method is specific for accurate diagnosis, the pragmatic inability to search thousands of diagnoses from patient medical records makes this methodology less sensitive. By contrast, the full medical record of cases identified within the Rochester Epidemiology Project medical records-linkage system was manually reviewed for the potential incident diagnosis of vestibular schwannoma. Although less specific, this approach allowed for greater sensitivity for the detection of vestibular schwannoma. Furthermore, despite the time-intensive review of patients’ full medical records, the relatively few numbers of potential cases of vestibular schwannoma that ultimately required manual review made this approach feasible. Since most benign brain tumors, like vestibular schwannoma, do not require pathological confirmation of their disease for diagnosis, the current work highlights how the reliance of large national brain tumor and cancer registries on pathology specimens and cancer-related treatment data to identify new diagnoses likely accounts for the discrepancies surrounding national estimates of disease incidence.

The unique identification of small tumors by the Rochester Epidemiology Project infrastructure helps to further understand the discrepancy in incidence rates. As a natural extension, patients with small tumors most commonly pursue a management modality consisting of observation with serial neuroimaging. Moreover, up to a quarter of all new diagnoses of vestibular schwannoma are made incidentally following the workup for unrelated conditions (7,14). Because these cases are comprised of exclusively “clinical diagnoses” without corroborating pathology or treatment data, they present the greatest challenge for detection under existing standardized protocols that facilitate the diagnosis of cancer, which require tissue specimens for diagnosis. Consequently, it is likely that some degree of systematic error surrounding the detection of small tumors in the medical record contributes to the lower incidence rates observed in large population-based brain tumor and cancer registries in the United States. Moreover, the rising incidence rates observed globally are largely driven by the increasing diagnosis of small tumors (7,8,19,20), and this likely explains why incidence rates reported from national brain tumor and cancer registries have been essentially stable since their inception in 2004. Taken together, these observations characterize a selection bias towards the registration of patients who have more advanced disease and therefore require treatment with either microsurgery or radiosurgery.

Importantly, these data characterize methodologic limitations surrounding the reporting of vestibular schwannoma that also likely extend to the diagnosis of other common benign brain tumors such as meningioma and pituitary adenoma. The limitations of these large brain tumor and cancer registries lead to particular underreporting of the diagnosis of small, oftentimes incidentally diagnosed, tumors that undergo observation with serial imaging. The underestimation of disease incidence and overestimation of disease severity may influence reported treatment outcomes, as well.

Of note, it is unlikely that true biological differences among countries or within the United States itself adequately account for the observed differences in disease incidence. First, incidence rates exceeding five per 100,000 person-years have been reported from distinct populations within the United States with varying genetic origins (6,7). Moreover, most published national and international data support that the large discrepancy in incidence rates stems from patients’ access to healthcare and neuroimaging (19,20,21). Additionally, population-based research from distinct populations, such as Taiwan and Denmark, have demonstrated incidence rates as high as 3.7 and 3.4 per 100,000 person-years in recent years, respectively. Chronicling the incidence rates from countries like Taiwan or Denmark over recent decades suggests that their detection rates are following about a decade behind the Rochester Epidemiology Project data, likely secondary to patients’ access to healthcare and frequency with which magnetic resonance imaging is used (8). A possibility remains that the low national incidence reported from brain tumor and cancer registries may be secondary to limited access to healthcare in the United States. Although this possibility cannot be entirely refuted by the current work, if true, its existence would not mean that the true national incidence of vestibular schwannoma is lower, but instead that national detection rates are considerably lower than other Asian and European countries, which appears unlikely given the United States leads the world in healthcare expenditures (22). Taken together, the incidence of vestibular schwannoma (and likely other benign brain tumors) is underreported compared with international data, and these differences are substantial: based on data from other countries, it is possible that twice as many people in the United States have been diagnosed with vestibular schwannoma compared with estimates from national brain tumor and cancer registries.

The primary limitation of the current work is that a single institution’s brain tumor and cancer registry is used as a surrogate for data obtained by numerous institutions included in the large population-based brain tumor and cancer registries within the United States. Even though our methodology of tracking cases will be similar and reporting of cases is standardized across the country, if our institution was particularly poor at identifying and registering cases of vestibular schwannoma, then the data would support an explanation for the discrepancy in reported incidence rates that is not representative of the true etiology. Although this limitation cannot be outright refuted, it is worth noting that there are multiple unique characteristics of our institution that, at the very least, suggest that it would provide a reasonable representation of institutions providing data to brain tumor and cancer registries that comprise these larger population-based cohorts from the United States. The mirroring incidence rates and clinical data from our institutional data and data reported from SEER or CBTRUS also support this assertion. Lastly, stemming from the serendipitous connection to the Rochester Epidemiology Project, the methodology employed in the current study is likely difficult to replicate elsewhere. Of note, an ostensible limitation of the current work is that the apparent discrepancy in incidence rates could be secondary to the diagnosis of vestibular schwannoma at institutions outside of Mayo Clinic within Olmsted County. These diagnoses would therefore be captured by the Rochester Epidemiology Project but not the institutional brain tumor and cancer registry. However, there is only one other primary medical center in Olmsted County, and fortuitously for the current investigation, every patient registered in the Rochester Epidemiology Project with vestibular schwannoma underwent neurotological and neurosurgical evaluation at Mayo Clinic upon diagnosis of their tumor. Lastly, the age and sex distributions of Olmsted County resemble the United States at large; however, 86% of Olmsted County self-reports as white versus only 72% of the national populace, and this could result in a slightly higher incidence of vestibular schwannoma in the Olmsted County population, although both the institutional and population-based incidence estimates would assumedly be equally affected by this bias (12).

CONCLUSION

Large cancer registries likely underreport the incidence of vestibular schwannoma in the United States. The reliance on pathology specimens and cancer-related treatment data for the national registration of new cancer and brain tumor diagnoses may introduce selection bias and underreporting of benign brain tumors that frequently involve observation as a primary treatment modality. This selection bias likely accounts for the discrepant incidence rates of vestibular schwannoma reported between the United States and other countries.