Rising incidence of sporadic vestibular schwannoma: True biological shift versus simply greater detection

Abstract

Objective:

The incidence of sporadic vestibular schwannoma (VS) has increased significantly over recent decades. The rising incidence of VS has been largely attributed to the increasing use of magnetic resonance imaging (MRI), especially with regard to incidentally diagnosed tumors. However, no study to date has directly investigated this supposed etiology beyond the observation that VS incidence rates have risen in the post-MRI era. Therefore, the primary objective of the current study was to characterize the incidence of head MRIs over the previous two decades in Olmsted County, Minnesota and compare this trend to the incidence of asymptomatic, incidentally diagnosed VS over the same time period.

Study Design:

Population-based incidence study.

Setting / Patients:

Using the unique resources of the Rochester Epidemiology Project, procedure codes for head MRIs and diagnostic codes for VS among residents of Olmsted County, Minnesota between 1 Jan 1995 and 31 Dec 2016 were retrieved. Incidence rates of head MRI and incidentally diagnosed VS were calculated on a per-year basis.

Results:

A total of 43,561 head MRIs among 30,002 distinct persons were identified from 1995 to 2016. The incidence of head MRI significantly increased between 1995 and 2003 (p<0.001), but remained stable between 2004 and 2016 (p=0.14). Over the same time interval, 25 cases of incidentally diagnosed VS were identified. The incidence of asymptomatic VS increased over time from 0.72 per 100,000 person-years between 1995 and 1999 to 1.29 between 2012 and 2016 (p=0.058). No plateauing of incidence rates was observed in incidental tumors over the study period. The size of incidentally diagnosed tumors did not change over the study period (p=0.93), suggesting that the increasing incidence of asymptomatic tumors is not explained by improved diagnostic capability of more recent MRI studies.

Conclusions:

Despite the plateauing of head MRI incidence rates after 2004, the incidence of asymptotic, incidentally diagnosed VS continued to increase. Our findings suggest that there may be additional contributory etiologies for the rising incidence of VS beyond greater detection alone.

INTRODUCTION

Multiple reports from Europe, Asia, and the United States have shown that the incidence of sporadic vestibular schwannoma (VS) has risen significantly over recent decades, approaching upwards of 5 per 100,000 person-years among all age groups and 20 per 100,000 person-years in persons aged 70 years and older (Table 1).^1–4 The most cited explanations for this rising incidence stem from the combination of improved disease awareness, more widespread adoption of screening protocols for asymmetrical sensorineural hearing loss (SNHL), and the increasing access to high-resolution head magnetic resonance imaging (MRI). Because these factors improve detection of VS, they are thought to represent the predominant driving force behind the rising incidence of sporadic tumors.

Table 1.

Modern incidence rates of sporadic vestibular schwannoma in Olmsted County, Minnesota between 2012 and 2016 by age at diagnosis.

Age, years	Total Population	Cases, N	Incidence*
0–19	211,380	0	0.0
20–29	111,045	0	0.0
30–39	110,091	0	0.0
40–49	90,176	3	3.3
50–59	106,437	12	11.3
60–69	73,409	10	13.6
≥70	72,925	15	20.6
All ages	775,463	40	5.2 (4.5†)

^*Incidence per 100,000 person-years.

^†Incidence per 100,000 person-years directly standardized by age and sex to the 2000 United States Census total population.

To a certain degree, this explanation is unequivocally contributing to the rising incidence of VS. Several studies have demonstrated heightened detection rates – and therefore higher incidence rates – among populations with greater access to healthcare.^2,4–6 For instance, the highest modern incidence rate reported in the literature to date comes from Beverly Hills, California at 5.4 per 100,000 person-years.⁵ Similarly, the incidental diagnosis of VS during the workup of other conditions necessarily contributes to rising incidence rates because the proportion of incidental diagnoses among all new diagnoses of sporadic VS is rising.^7,8 For instance, from 1995 to 1999 in Olmsted County, Minnesota, 20% of new diagnoses of VS were incidentally diagnosed, but from 2012 to 2016, incidental tumors comprised 28% of new diagnoses.^2,9

At the same time, true shifts in disease epidemiology are not uncommon. For example, within the field of otolaryngology, in the last 30 years we have seen a dramatic rise in the incidence of oropharyngeal carcinoma while also observing a significant decline in the incidence of otosclerosis.^10,11 Therefore, a true biological shift in the epidemiology of VS would not be unprecedented. To this end, the degree to which greater detection alone influences the observed rising incidence of VS is unknown. Beyond the recognition that the incidence of VS has risen in the post-MRI era, the question regarding the degree to which greater detection contributes to the rising incidence has been left largely unanswered.

The current study was conceived with the chief objective of comparing the incidence trends in head MRI and the incidental diagnosis of sporadic VS over time in Olmsted County, Minnesota. Of note, analysis of the trends in incidentally diagnosed tumors is fundamental to addressing the question surrounding the role greater detection plays in the rising incidence of VS because trends in symptomatic tumors are necessarily influenced by heightened disease awareness and screening protocols for asymmetrical SNHL. Thus, theoretically, comparing the trends in incidence rates of asymptomatic, incidentally diagnosed tumors to the rate of head MRI studies in a population should provide insight into the influence that detection alone plays.

METHODS

Clinical Data

Following approval from the Mayo Clinic and Olmsted County Medical Center Institutional Review Boards (IRB 15–006036 and 050-OMC-15, respectively), procedure codes for head MRIs performed on residents of Olmsted County, Minnesota between 1 Jan 1995 and 31 Dec 2016 were retrieved using the unique resources of the Rochester Epidemiology Project (REP). Since 1966, virtually all residents of Olmsted County, Minnesota have been prospectively registered with the REP either at birth or on the date of their first healthcare visit as a resident of Olmsted County. Multiple studies have validated the accuracy of the REP, and the rate of under-inclusion of 1 or more records from a given patient’s chart is estimated to be approximately 1%.¹² Procedural codes for MRIs began being documented in 1995 (CPT codes 70551, 70552, and 70553), and we used this as the inception date of our study. The methodology behind the retrieval and review of all cases of VS among residents of Olmsted County, Minnesota has been described previously.^2,9

Statistical Methods

Continuous features were summarized with medians, interquartile ranges (IQRs), and ranges. Trends in tumor size over time were evaluated using Spearman rank correlation coefficients. Incidence rates per 100,000 person-years were calculated using incident cases of VS or head MRIs for each calendar year from 1995 to 2016 as the numerator and age- and sex-specific counts of the population of Olmsted County, Minnesota as the denominator. The denominators were obtained from a complete enumeration of the Olmsted County population provided by the REP.¹³ Only the first MRI per person per year was included in the analysis for persons with more than one MRI. Incidence rates were directly standardized to the total United States population from the 2000 United States Decennial Census. Associations with incidence rates were evaluated using Poisson rate regression models. Statistical analyses were performed using version 9.4 of the SAS software package (SAS Institute; Cary, NC), and p-values <0.05 were considered statistically significant.

RESULTS

Incidence of Head MRIs Since 1995

We identified 43,561 head MRIs among 30,002 distinct persons. The 43,561 head MRIs included 25,174 among women and 18,387 among men with a median age of 51 years (IQR 33–68; range 0–103). The median number of MRIs contributed per person during the study was 1 (IQR 1–2; range 1–22). The age-adjusted incidence rate of head MRI was 1,546 per 100,000 person-years for women and 1,310 per 100,000 person-years for men. Incidence rates by year from a model with a smoothing spline for year are shown in Figure 1A, suggesting that incidence rates of head MRI increased from 1995 to 2003, but leveled off from 2004 to 2016. Based on a log-linear model separated by the start of the plateauing period in 2004, the incidence of head MRI was higher among women compared to men (p<0.001), increased from 1995 to 2003 (9.6% increase per year; p<0.001), and remained fairly constant from 2004 to 2016 (0.2% increase per year; p=0.14).

Figure 1.

Age-adjusted incidence rates by year from a model with a smoothing spline showing the incidence of (A) head MRI, (B) asymptomatic, incidentally diagnosed vestibular schwannoma (VS), and (C) symptomatic VS.

Incidence of Asymptomatic, Incidentally Diagnosed VS

Twenty-five incident cases of asymptomatic, incidentally diagnosed VS were identified, including 17 women and 8 men, with a median age at diagnosis of 62 years (IQR 53–70; range 30–82). The age-adjusted incidence rate of asymptomatic VS was 1.03 per 100,000 person-years for women and 0.55 per 100,000 person-years for men. Incidence rates by year from a model with a smoothing spline for year are shown in Figure 1B. Based on a log-linear model, the incidence of asymptomatic VS was higher among women compared to men (p=0.11) and increased during the study timeframe (6.3% increase per year; p=0.058). For example, the age- and sex-adjusted incidence of asymptomatic VS increased from 0.72 per 100,000 person-years between 1995 and 1999 to 1.29 between 2012 and 2016. Tumor size did not appreciably change over the study period (p=0.93), nor did it change following the observed plateau in head MRI incidence after 2004 (p=0.82). Lastly, after 2012, nearly all head MRIs performed at Mayo Clinic were done using 3T MRI with an average slice thickness of 2 mm. However, despite 75% of the tumors post-2012 being diagnosed on a 3T scanner, the tumor size did not appreciably change compared to tumors diagnosed before 2012 (p=0.89).

Symptomatic VS

Ninety-five incident cases of symptomatic VS were identified, including 46 women and 49 men, with a median age at diagnosis of 59 years (IQR 50–69; range 20–86). The age-adjusted incidence rate of symptomatic VS was 2.74 per 100,000 person-years for women and 3.52 per 100,000 person-years for men. Incidence rates by year from a model with a smoothing spline for year are shown in Figure 1C. Based on a log-linear model, incidence of symptomatic VS was higher among men compared to women (p=0.30) and increased during the study timeframe (2.2% increase per year; p=0.19), although neither trend reached statistical significance.

DISCUSSION

The rising incidence of sporadic VS has been largely attributed to heightened detection secondary to improved disease awareness and widespread use of MRI. However, beyond the observation that incidence rates have risen in the post-MRI era and the intuitive conclusions surrounding the increasingly frequent diagnosis of incidental tumors, the degree to which greater detection alone influences the rising incidence of VS is unknown. Furthermore, the potential of a true biological shift in the epidemiology of a disease over recent years should not be considered far-fetched. Just within the last 30 years, the field of otolaryngology has witnessed dramatic changes in the epidemiology of both oropharyngeal carcinoma and otosclerosis, the former because of an established viral driver of disease pathogenesis and the latter because of a yet unknown etiology.^10,11

The current study attempts to quantify the influence that greater detection alone has on the rising incidence of VS by comparing the trends of incidence rates of head MRIs and VS, in particular asymptomatic, incidentally diagnosed VS, in the Olmsted County population. From 1995 to 2004, the incidence of head MRIs increased significantly; however, after 2004 the incidence plateaued and remained stable through 2016. By contrast, the incidence of incidentally diagnosed VS steadily rose at a rate of 6.3% per year and no plateau was observed. Moreover, the tumor size of incidentally diagnosed VSs did not change over the study period. This suggests that, despite the plateaued incidence of head MRI, improved diagnostic capabilities of MRI are not resulting in the detection of smaller incidental tumors over time. Furthermore, despite a practice shift at Mayo Clinic towards performing nearly all head MRIs on a 3T scanner after 2012, the tumor size at diagnosis did not appreciably change. These observations support the conclusion that the resolution of diagnostic MR technology for tumor detection has not improved substantially since 2004. The observed trends in the incidence of incidentally diagnosed tumors are also not simply secondary to an aging population. Two features of the current study support this: First, the reported incidence rates were age-adjusted, and second, the median age of patients incidentally diagnosed with VS is comparable to other population-based studies on disease epidemiology.^1,2 Together, this suggests that the rising incidence of incidentally diagnosed tumors is, at least in part, independent of trends in the incidence of head MRI. Lastly, the trend in incidence of symptomatic tumors slightly increased over the study period, eventually plateauing around 2006. This suggests that the influence greater detection bears on the modern incidence of VS through improved disease awareness and screening protocols for asymmetrical SNHL has stabilized. This also suggests that further increases in the incidence of VS will expectedly be secondary to incidentally diagnosed cases. Taken together, although certainly influenced by greater detection, greater detection alone cannot adequately account for the sustained rising incidence of incidentally diagnosed tumors. Therefore, the data from the current study suggest that a true biological shift could also be contributing to rising incidence rates of VS.

Several studies have investigated possible mechanisms that could be contributing to the rising incidence of VS. The most researched area surrounds potential associations between environmental exposures and the development of VS;^14–16 however, the data supporting associations between these exposures and the sporadic development of VS are weak. For instance, the studies showing associations between cell phone use or noise exposure and the development of VS are all survey-based, case-control studies.^14–16 One of the most well-established limitations of this type of study design is the susceptibility of the data to display recall bias, where the “cases” exhibit an increased likelihood to report a positive exposure history compared to the “controls” due to their awareness of their disease. In the case of VS research, this is likely a major factor behind the positive associations: a patient who has been counseled that they have developed a tumor on their hearing nerve is more likely to report exposures related to hearing. Similarly, a person who frequently uses a cell phone is more likely to notice asymmetrical hearing loss and undergo head imaging. Substantiating the limitations of past case-control studies, no significant associations between these exposures and the development of VS were detected by non-survey, population-based studies.^17,18 Therefore, if a true biological shift exists underlying the rising incidence of VS, an attributable etiology has yet to be discovered. However, it is worth noting that the absence of such an etiology would not be unique. For instance, the incidence of otosclerosis has decreased over recent decades for unknown reasons.¹⁹

Lastly, there are several limitations of the current work. First, the number of incidentally diagnosed VS cases in our population over the study period is small (25 incident cases). Therefore, more data are needed to determine if this trend persists or if the trends converge with time. A second limitation surrounds the generalizability of findings in the Olmsted County population. We have reported previously that Olmsted County exhibits higher incidence rates of VS than the populations in most other epidemiologic studies. We suspect that this is driven by the area’s unique access to healthcare and, in particular, the fact that nearly one-third of persons aged 70 years and older living in Olmsted County have had a head MRI.^2,9 If this rationale is correct, these data should not be taken to mean that the incidence of VS is truly higher in Olmsted County – instead, they would indicate that the incidence is likely similarly high across the United States, barring differences in demographic features, and the healthcare infrastructure of Olmsted County simply detects what is already present at a higher rate. Furthermore, we also note that the age, sex, and race/ethnicity characteristics of Olmsted County have been shown to be similar to a large portion of the Upper Midwest US population.²⁰

CONCLUSION

Despite the plateauing of head MRI incidence rates after 2004, the incidence of asymptotic, incidentally diagnosed VS continues to increase. Our findings suggest that there may be additional contributory etiologies for the rising incidence of VS beyond greater detection alone.