FIG. 1.
Long-term HRQOL outcomes using the disease-specific PANQOL scale among 103 nontumor controls and 539 patients with VSs smaller than 3 cm who were managed with SRS (n = 247), observation (n = 148), and microsurgery (n = 144). A higher score denotes better HRQOL.
Error bars represent 95% confidence intervals for the group score average. Figure is available in color online only.
Multivariate comparisons of HRQOL outcomes of nontumor controls and patients who underwent SRS, observation, and microsurgery are shown in. Additionally, a comparison between nontumor controls and the collective population of patients with VS is provided. Of note, after adjusting for age, the PROMIS physical, and the SF-36 Physical and Mental Component Summary scores did not show any statistically significant differences between nontumor control subjects and any of the individual management arms. Compared with the nontumor control group, patients with VS consistently had poorer PANQOL facial, PANQOL balance, PANQOL hearing, and PANQOL total scores. As a general rule, the mean difference in scores between control subjects and patients with VS was greater than any of the differences seen between individual management groups.
.
DiscussionOver the last century, the treatment of VS has undergone tremendous refinement resulting from improvements in surgical technique, advancements in radiation delivery, and a better understanding of the natural history of disease. We have now reached a relative plateau with several important objective clinical outcome measures. Namely, more than 90% of patients receive durable tumor control and satisfactory facial nerve function, and disease-specific mortality remains exceedingly low whether treated with microsurgery or SRS. While microsurgical resection remains the preferred treatment for tumors larger than 3 cm, management practices for small- to medium-sized tumors vary significantly between treatment centers, reflecting the relative lack of evidence supporting one therapy over another. Despite the large number of studies evaluating VS outcomes, there remains no consensus regarding optimal management.
Unfortunately, it is not uncommon that we, as health care providers, encounter a patient dissatisfied with his or her outcome despite having what we perceive to be an excellent treatment result. Over the past several decades there has been increasing importance placed on patientcentered outcomes, realizing that clinicians commonly underestimate patient impairment, and there is frequent disparity between what health care providers value and what patients prioritize. Additionally, treating physicians may not fully appreciate the psychological impact of the diagnosis of a brain tumor, even a benign one, and the significant impact it may have on HRQOL even prior to embarking on a plan of care. Outcome assessment tools developed by the health care industry often evaluate a narrow range of technical outcomes and frequently fail to provide insight into how the disease itself or the untoward effects of treatment influence overall patient wellbeing. With an increasing prevalence of small tumors in minimally symptomatic patients it is paramount that we carefully examine both traditional clinical measures and HRQOL data when determining best practices.
Previous studies evaluating HRQOL in VS have carried several notable limitations including low patient numbers, lack of a normal non-VS control population, exclusion of one or more treatment groups, and short-term follow-up and have therefore been criticized for being methodologically weak. To date, only 3 prospective, nonrandomized studies have been published comparing HRQOL outcomes following treatment, and currently no Level 1 evidence exists. It is unlikely that a prospective randomized trial comparing treatment outcomes will ever materialize given a significant number of obstacles, including patient recruitment in a relatively uncommon condition, the number of subjects required to detect clinically meaningful differences, and significant practice disparities among many major centers, making multicenter collaborations difficult. In fact, Myrseth et al. had to abandon their initial plan to randomize enrollment, as patients were unwilling to relegate their treatment to chance.
22Pollock et al. reported the first prospective nonrandomized study in 2006 including 82 patients with primary sporadic, smaller than 3-cm VSs who underwent microsurgery or SRS. The mean duration of follow-up was 42 months, and the Health Status Questionnaire (modification of the SF-36) was used to assess longitudinal change in HRQOL. The SRS cohort did not experience any deterioration in HRQOL, while the microsurgical arm saw an early decline in the standardized physical component score at 3 months following treatment, which subsequently returned to near baseline. At last follow-up the net change from baseline for the physical and mental summary scores for both treatment groups were very similar; there was a net loss of 3.1 and 3.6 points in the standardized physical component score and a net improvement of 4.5 and 3.7 points in the standardized mental component score for the SRS and microsurgery groups, respectively.
Subsequently, in 2009 Myrseth and colleagues published the second prospective nonrandomized study comparing SRS and microsurgery in a cohort of 91 Norwegian patients using the SF-36 and GBI to evaluate longitudinal changes in HRQOL at baseline, 1 year, and 2 years following treatment. At last follow-up, there were no statistical differences between treatment groups on individual or summary scores for the SF-36. Similarly, using the GBI, there were no differences between groups at 1 year after treatment. However, at 2 years there was a statistically significant decline in the GBI total score, and the general and physical subdomain scores for the microsurgery arm. At last follow-up, there were no statistically significant differences in job disability or sick leave status between groups.
Di Maio et al. published the most recent prospective study including 47 patients who underwent observation, 48 who received linear accelerator–based radiotherapy, and 97 who underwent microsurgical resection of VS smaller than 3 cm in size. At a mean follow-up of 31.8 months, there were no statistically significant differences between groups on either the SF-36 physical or mental health summary score; all 3 groups were within approximately 4 points of baseline values. Notably, HRQOL remained unchanged for the observation and radiotherapy cohorts throughout follow-up, while the surgical group experienced a significant improvement in mental health at 24 months, which subsequently returned to baseline at last follow-up.
In the current study, patients who underwent microsurgery had statistically significantly worse outcomes in their PANQOL total score as well as the subdomains of facial function, balance, and pain compared with subjects treated with SRS or observation. Considering that the incidence of objective facial nerve dysfunction following microsurgery was approximately 10% greater than SRS or observation, it is not surprising that these patients had a statistically worse facial PANQOL score.
Furthermore, as primary VSs are rarely associated with trigeminal neuralgia or other pain syndromes and since prolonged headache following surgery is a known potential complication, it is not unexpected that the microsurgery cohort had an average poorer score than the SRS or observation groups in the pain subdomain as well. Several studies have demonstrated that ongoing dizziness, compared with other variables such as facial nerve dysfunction and hearing loss, has the most profound effect on HRQOL. It is interesting that the microsurgery cohort scored poorer in the PANQOL balance subdomain since it is commonly held that tumor resection is advantageous for treatment of patients who are experiencing debilitating vertigo. This seemingly paradoxical observation can be explained by the differences in dizzy symptoms patients experience following microsurgery, SRS, and observation as well as treatment selection bias. Microsurgery nearly always imparts a complete ipsilateral peripheral vestibular paresis, often resulting in acute postoperative vertigo; however, over a period of several weeks or months most subjects achieve central compensation with a significant improvement in symptoms. In contrast, patients who are observed or receive SRS often have residual ipsilateral vestibular function, which reduces the risk of mild symptoms associated with unilateral vestibular hypofunction (e.g., transient imbalance with quick head turns), but with a functioning labyrinth, patients may still experience incapacitating vertigo. Since the prevalence of longterm disabling vertigo is low, by shear number, the milder symptoms of unilateral vestibular hypofunction seen in the surgical cohort dominates the PANQOL balance score rather than the few outlier patients with debilitating vertigo following SRS or observation. This may also be explained by treatment selection bias. In our practices, patients with growing tumors and incapacitating vertigo are more likely to receive microsurgery than SRS or observation. This strategy may inherently select for those that are more sensitive to dizziness, which will likely negatively influence balance scores following treatment.
There are several important considerations that should be taken into account when interpreting VS HRQOL data. First, it is generally held that microsurgery imparts upfront morbidity with subsequent improvement over time, while the untoward effects of radiation and observation are often not evident for many years following treatment; therefore, long-term analysis is critical to ensure fair comparison. To illustrate this point, Pollock et al. demonstrated that at 3 months after treatment, patients who underwent surgery had a statistically significant decline in several of the SF-36 domains, including physical functioning, role-physical, energy/fatigue, and overall physical. However, at last follow-up only a decline in bodily pain persisted. Furthermore, since most patients are diagnosed in their 4th to 5th decade of life and are expected to live 20 or more years beyond diagnosis, late HRQOL data should be prioritized over transitory HRQOL changes found early after treatment.
One other important consideration when reviewing HRQOL data is whether small statistically significant differences between groups are clinically meaningful. That is to say, is a several point discrepancy even perceptible to the patient or important to the clinician? For example, although statistically significant, the difference between observation and SRS on the PANQOL total score was only 2 points on a 100-point scale, and there were only 5 points separating SRS and microsurgery. The minimum clinically important difference (MCID) was developed to assist in answering this complex but important question. The MCID represents the minimum change in score that a patient would perceive as an improvement over baseline for a particular condition. While the MCID of the SF-36 has not been specifically evaluated for patients with VS, the MCID for many other studied disease conditions is commonly greater than 5 points for the physical and mental health summary scores. Notably, none of the 3 prospective studies discussed earlier exceeded a mean 5 point change in either score at last follow-up.
Until very recently, one significant obstacle to HRQOL assessment has been the lack of a validated disease-specific instrument. By default, the SF-36 has become the most widely used instrument for assessing HRQOL outcomes within the VS literature. Importantly, the SF-36, GBI, and PROMIS-10 are multipurpose instruments and therefore are heavily influenced by patient comorbidity and may lack the sensitivity required to detect subtle changes after treatment or small differences between management arms.
4 In 2010 Shaffer and colleagues introduced the disease-specific PANQOL scale, a 26-item questionnaire evaluating facial function, balance, hearing, pain, anxiety, energy, and general health. Initial evaluation yielded robust face, content, criterion, and construct validity with high test-retest reliability. The current study corroborates previous data demonstrating that the PANQOL is a more responsive instrument to disease- and treatment-associated changes than multipurpose measures such as the SF-36 or PROMIS-10. Specifically, after multivariable adjustment, the latter 2 instruments did not demonstrate any statistically significant differences between patients receiving SRS, observation, or microsurgery. Moreover, in almost every case the SF-36 and PROMIS-10 failed to detect differences between nontumor controls and patients with VS, while the PANQOL did so in almost every subdomain score. Moving forward, we believe that the PANQOL scale should supplant the SF-36 as the default instrument for VS HRQOL assessment. The next important step will be to estimate the MCID for the PANQOL total and subdomain scores from prospective analysis.
Since intervention does not offer an HRQOL advantage over observation, and because a significant number of small- to medium-sized VSs do not grow after diagnosis, tumors should be initially observed using serial imaging. We believe that it is hard to justify the added cost and potential risks, albeit small, of upfront SRS without evidence to support an HRQOL advantage over the natural history. If growth is documented on serial imaging, or additional tumor-related symptoms develop, active treatment with either SRS or microsurgery is indicated. For patients with tumor growth and minimal symptoms, SRS appears to offer the best HRQOL outcome, as long as long-term tumor control is achieved. Finally, microsurgery should be considered for patients with mass effect symptoms, growing cystic tumors, or strong preference.
In the current study, PANQOL scores were reduced in patients with VS compared with general population nontumor controls; however, the differences between individual management groups at 8 years following treatment were relatively small. These data support the notion that the diagnosis of VS imparts a significant reduction on HRQOL, while long-term differences between treatment strategies are generally less significant by comparison. We believe that further meaningful gains in HRQOL will require a redirection of efforts. While facial nerve function and tumor control are important, results from large-volume centers have largely plateaued within the last several decades. Moving forward, improvements in patient counseling regarding realistic long-term expectations of disease, as well as offering pre- and posttreatment psychosocial support, may provide the most impact on HRQOL improvement for patients with VS.
For example, it is intriguing that there was no difference between groups in the PANQOL hearing domain despite patients undergoing SRS and microsurgery having significantly higher rates of severe hearing loss following treatment. This very well could be explained by the fact that we extensively counsel patients about the risks and consequences of unilateral deafness at initial consultation and during follow-up.
We wish to highlight some strengths of the current study and finally acknowledge several limitations. This is the first multicenter study evaluating HRQOL in VS and includes the largest number of patients to date. The combination of a high questionnaire response rate and international multicenter participation permitted enrollment of a large number of subjects, improving statistical inference, and enhancing external validity, and, therefore, the generalizability of our findings. Second, the majority of high-volume tertiary centers carry a strong preference for either SRS or microsurgery, and, consequently, most studies exclude one or more treatment arms. In the current study, both centers have significant experience managing VS using SRS and microsurgery with a relatively even distribution between treatment strategies.
Additionally, a significant number of small- to medium-sized tumors are conservatively followed with serial MRI and audiograms at biannual or annual visits. As a result, we were able to compare HRQOL outcomes between all 3 management arms, in addition to a group of nontumor control subjects. Third, all patients were surveyed using 3 general HRQOL instruments and a recently validated disease-specific questionnaire to minimize the risk of monomethod bias present in earlier studies. To our knowledge, the PROMIS-10 has not been previously used in VS HRQOL assessment. Finally, long-term HRQOL analysis is critical to evaluating differences between microsurgery, SRS, and observation. The mean time interval between treatment and survey was 7.7 years—more than twice the follow-up of previous studies.
The primary limitations of the current study are nonrandomized treatment and the lack of longitudinal follow-up. Without patient randomization to observation, SRS, or microsurgery, it remains possible that treatment selection bias may result in baseline differences between groups, which may in turn influence long-term HRQOL outcome results. For example, patients who have resigned themselves to the diagnoses and are able to cope with disease-related symptoms are more likely to be observed or receive SRS, while those who are overly anxious about the diagnosis of a brain tumor or are most bothered by what are felt to be tumor-related symptoms, such as trigeminal neuropathy or recurrent vertigo, are more likely to receive microsurgery. Therefore, it is possible that the slightly lower scores in the microsurgery arm are the result of pretreatment patient-related factors rather than treatment. Furthermore, without pretreatment HRQOL scores and longitudinal follow-up, time-dependent changes in HRQOL and causality between intervention and outcome cannot be confidently established. Finally, both centers in the current study are tertiary academic hospitals, and therefore it is possible that the VS population is biased by referral patterns. Once again, while considered the gold standard, randomized controlled trials comparing management strategies may never materialize given the relative rarity of disease, reluctance of patients to resign control over choice of management, strong biases of treating physicians, and the large number of patients that would be required to detect meaningful differences.
ConclusionsOverall, the HRQOL differences following observation, microsurgery, and SRS are small. The diagnosis of VS, rather than treatment strategy, most significantly impacts patient quality of life. Therefore, further improvements in patient counseling regarding realistic long-term expectations of disease, as well as offering pre- and posttreatment psychosocial support, may provide the most impact on HRQOL improvement for patients with VS.
Since a significant number of tumors do not grow after the time of diagnosis, and because intervention does not appear to confer an HRQOL advantage over observation, small- and medium-sized VSs should be initially observed, while active treatment should be reserved for patients with unequivocal tumor growth or intractable symptoms that are amenable to treatment. As disease-specific HRQOL outcomes following SRS appear to be minimally better than microsurgery, we believe that SRS should be offered to those with growing tumors and minimal symptoms, while microsurgery should be considered for patients with symptoms associated with mass effect, enlarging cystic tumors, or in cases of patient preference. Moving forward, studies assessing HRQOL in VS patients should prioritize use of validated disease-specific measures, such as the PANQOL, given the significant limitations of generic instruments in distinguishing between treatment groups and tumor versus nontumor subjects.