Abstract
Purpose
The aim of this study is to rigorously evaluate the role of auditory brainstem response (ABR) testing in the diagnosis of vestibular schwannomas (VS).
Materials and methods
Searches were conducted in multiple online databases, supplemented by hand searches. From the studies chosen for final inclusion, relevant data were extracted and meta-analysis of pooled data was performed.
Results
623 studies were identified from which 43 met inclusion criteria for analysis (1978 to 2009) including 3314 patients. Pooled sensitivity for ABR detection of vestibular schwannomas was 93.4% (95% CI 92.6–94.3, P = 0.0000). For tumors less than 1 cm (8 studies, 176 patients) sensitivity was 85.8% (95% CI 80.6–90.1, P = 0.0116). For tumors greater than 1 cm (6 studies, 251 patients) pooled sensitivity was 95.6% (95% CI 93.1–98.2, P = 0.0660). Sensitivity of ABR to detect extracanalicular tumors was higher than for intracanalicular tumors, though pooled data were not statistically valid. Pooled specificity (8 studies, 2432 patients) was 82.0% (95% CI 80.5–83.6, P = 0.0000).
Conclusions
Although MRI remains the gold standard, emerging trends towards more conservative management coupled with limited financial resources may prompt many clinicians to review the role of ABR testing in screening for retrocochlear pathology. In light of the high sensitivity and specificity of ABR testing for VS, we strongly urge its reconsideration as a useful diagnostic tool for patients with clinically suspected VS.
1
Introduction
In 1977, Selters and Brackmann first demonstrated how the auditory brainstem response (ABR) could serve as a useful tool for detecting vestibular schwannomas (VS), and shortly thereafter the ABR became regarded as the initial diagnostic test of choice in patients with clinically suspected VS . However, with improvements in contrast-enhanced magnetic resonance imaging (MRI) in the late 1980s, the usefulness of the ABR in the diagnostic workup of VS was called into question, and currently MRI is widely considered the new gold standard for diagnosing retrocochlear pathology. Compared with ABR, MRI was shown to be more sensitive in detecting small tumors, and therefore theoretically more likely to result in improved surgical outcomes and hearing preservations rates .
However, management styles of vestibular schwannomas have changed considerably since the introduction and widespread use of MRIs. Current trends towards more conservative treatment paradigms, including observation (wait-and-scan), are based largely on improved understanding of the natural course of these benign growths. This, coupled with certain financial realities of limited healthcare resources, has prompted some practitioners to question the necessity of MRI for every patient with asymmetric hearing loss . With this in mind, the time may have come to revisit the potential role of ABR in diagnosing VS using meta-analysis.
A meta-analysis is used to synthesize data from numerous sources (often obtained from systematic review) resulting in pooled-data that can be analyzed for effect size. The combined data of numerous smaller studies can be used to test original hypotheses or generate new ones . The sensitivity and specificity of ABR in the detection of VS lend itself ideally to this kind of statistical analysis.
2
Materials and methods
2.1
Literature review
The search for relevant publications was conducted in accordance with current recommendations . Online searches were conducted in PubMed, CINAHL, Biosis, Cochrane, Cambridge Scientific Abstracts, Web of Science, and EMBASE for English-language studies that met search criteria. The references of relevant studies were also examined for additional candidates, and hand searches were performed in texts and journals.
The search terms ( Table 1 ) were used in all possible combinations to identify relevant studies. Returned studies were imported into RefWorks (RefWorks, Bethesda, Maryland) and duplicates were deleted. Two independent reviewers performed a secondary search using the inclusion and exclusion criteria listed in Table 2 . Data were extracted from studies that met criteria including author(s) and date of study, criteria for abnormal ABR, overall sensitivity, overall specificity, sensitivity based on tumor size (if available), and sensitivity data based on tumor location (if available).
| Auditory brainstem response | Vestibular nerve tumour |
| ABR | Vestibular nerve lesion |
| Brainstem evoked response | Acoustic neuroma |
| BAER | Acoustic neurinoma |
| BSER | Acoustic tumor |
| Electric response audiometry | Intracranial tumor |
| ERA | Intracranial tumour |
| Auditory steady state response | Intracranial lesion |
| ASSR | Cerebellopontine angle tumor |
| Vestibular schannoma | Cerebellopontine angle tumour |
| Vestibular nerve tumor | Cerebellopontine angle lesion |
| Inclusion Criteria | Exclusion Criteria |
|---|---|
| Patient in series underwent ABR screening for VS | Patients in series had previous CPA surgery |
| Criteria for abnormal ABR were reported | Patients had known CPA masses |
| Presence of surgically or radiographically confirmed VS | VS was not surgically confirmed |
| Sensitivity and specificity were reported or could be calculated from available data | Sensitivity and specificity were not reported or could not be calculated |
| Study passed quality assessment | Study did not pass quality assessment |
2.2
Statistical analysis
After extraction of necessary data, meta-analysis was performed for overall sensitivity and specificity, sensitivity based on tumor size, and sensitivity based on tumor location. Both fixed and random-effects models were used. The Q test for heterogeneity and I-squared statistic were calculated to identify possible outliers. Results for sensitivity and specificity of individual studies were also presented graphically to assess homogeneity.
2
Materials and methods
2.1
Literature review
The search for relevant publications was conducted in accordance with current recommendations . Online searches were conducted in PubMed, CINAHL, Biosis, Cochrane, Cambridge Scientific Abstracts, Web of Science, and EMBASE for English-language studies that met search criteria. The references of relevant studies were also examined for additional candidates, and hand searches were performed in texts and journals.
The search terms ( Table 1 ) were used in all possible combinations to identify relevant studies. Returned studies were imported into RefWorks (RefWorks, Bethesda, Maryland) and duplicates were deleted. Two independent reviewers performed a secondary search using the inclusion and exclusion criteria listed in Table 2 . Data were extracted from studies that met criteria including author(s) and date of study, criteria for abnormal ABR, overall sensitivity, overall specificity, sensitivity based on tumor size (if available), and sensitivity data based on tumor location (if available).
| Auditory brainstem response | Vestibular nerve tumour |
| ABR | Vestibular nerve lesion |
| Brainstem evoked response | Acoustic neuroma |
| BAER | Acoustic neurinoma |
| BSER | Acoustic tumor |
| Electric response audiometry | Intracranial tumor |
| ERA | Intracranial tumour |
| Auditory steady state response | Intracranial lesion |
| ASSR | Cerebellopontine angle tumor |
| Vestibular schannoma | Cerebellopontine angle tumour |
| Vestibular nerve tumor | Cerebellopontine angle lesion |
| Inclusion Criteria | Exclusion Criteria |
|---|---|
| Patient in series underwent ABR screening for VS | Patients in series had previous CPA surgery |
| Criteria for abnormal ABR were reported | Patients had known CPA masses |
| Presence of surgically or radiographically confirmed VS | VS was not surgically confirmed |
| Sensitivity and specificity were reported or could be calculated from available data | Sensitivity and specificity were not reported or could not be calculated |
| Study passed quality assessment | Study did not pass quality assessment |
2.2
Statistical analysis
After extraction of necessary data, meta-analysis was performed for overall sensitivity and specificity, sensitivity based on tumor size, and sensitivity based on tumor location. Both fixed and random-effects models were used. The Q test for heterogeneity and I-squared statistic were calculated to identify possible outliers. Results for sensitivity and specificity of individual studies were also presented graphically to assess homogeneity.
3
Results
The preliminary data search returned 623 studies excluding duplicates. From this initial pool 87 articles appropriate for secondary review were identified based on title and abstract. After performing secondary review, 43 studies were deemed appropriate for inclusion in the final study (1978 to 2009) including 3314 patients ( Fig. 1 , Appendix A).
Pooled sensitivity for ABR detection of vestibular schwannomas was 93.4 (95% confidence interval [CI] 92.6 to 94.3, P = 0.0000) ( Table 3 , Fig. 2 ). Pooled specificity (8 studies, 2432 patients) was 82.0% (95% CI 80.5–83.6, P = 0.0000) ( Table 4 , Fig. 3 ). For tumors less than 1 cm (8 studies, 176 patients) sensitivity was 85.8% (95% CI 80.6–90.1, P = 0.0116) ( Table 5 , Fig. 4 A ). For tumors greater than 1 cm (6 studies, 251 patients) pooled sensitivity was 95.6% (95% CI 93.1–98.2, P = 0.0660) ( Table 6 , Fig. 4 B). Sensitivity of ABR to detect extracanalicular tumors was higher than for intracanalicular tumors, though pooled data were not statistically valid ( Tables 7, Table 8 , Fig. 5 ).
| Study | TP | TP + FN | Sensitivity | SE (Sensitivity) | 95% CI Lower | 95% CI Upper |
|---|---|---|---|---|---|---|
| Thomsen, 1978 | 26 | 27 | 0.9630 | 0.0475 | 0.8699 | 1.0561 |
| Clemis, 1979 | 25 | 27 | 0.9259 | 0.0475 | 0.8328 | 1.0190 |
| Glasscock, 1979 | 48 | 49 | 0.9796 | 0.0353 | 0.9105 | 1.0487 |
| Eggermont, 1980 | 34 | 36 | 0.9444 | 0.0411 | 0.8638 | 1.0251 |
| Josey, 1980 | 51 | 52 | 0.9808 | 0.0342 | 0.9137 | 1.0479 |
| Djupesland, 1981 | 15 | 16 | 0.9375 | 0.0617 | 0.8166 | 1.0584 |
| Harner, 1981 | 25 | 26 | 0.9615 | 0.0484 | 0.8667 | 1.0564 |
| Rosenhall, 1981 | 30 | 30 | 1.0000 | 0.0451 | 0.9117 | 1.0883 |
| Terkildsen, 1981 | 55 | 56 | 0.9821 | 0.0330 | 0.9175 | 1.0468 |
| Bauch, 1982 | 25 | 26 | 0.9615 | 0.0484 | 0.8667 | 1.0564 |
| Maurer, 1982 | 37 | 37 | 1.0000 | 0.0406 | 0.9205 | 1.0795 |
| Bergenius, 1983 | 21 | 21 | 1.0000 | 0.0539 | 0.8944 | 1.1056 |
| Cashman, 1983 | 35 | 35 | 1.0000 | 0.0417 | 0.9182 | 1.0818 |
| Prasher, 1983 | 43 | 43 | 1.0000 | 0.0376 | 0.9262 | 1.0738 |
| Barrs, 1985 | 224 | 229 | 0.9782 | 0.0163 | 0.9462 | 1.0101 |
| Mangham, 1987 | 48 | 51 | 0.9412 | 0.0346 | 0.8734 | 1.0089 |
| Zollner, 1987 | 31 | 33 | 0.9394 | 0.0430 | 0.8552 | 1.0236 |
| Cohen, 1988 | 28 | 28 | 1.0000 | 0.0466 | 0.9086 | 1.0914 |
| Josey, 1988 | 90 | 93 | 0.9677 | 0.0256 | 0.9176 | 1.0179 |
| Telian, 1989 | 91 | 93 | 0.9785 | 0.0256 | 0.9283 | 1.0287 |
| Welling, 1990 | 63 | 68 | 0.9265 | 0.0299 | 0.8678 | 0.9851 |
| Grabel, 1991 | 51 | 56 | 0.9107 | 0.0330 | 0.8461 | 0.9754 |
| Levine, 1991 | 24 | 27 | 0.8889 | 0.0475 | 0.7958 | 0.9820 |
| Kotlarz, 1992 | 8 | 8 | 1.0000 | 0.0873 | 0.8290 | 1.1710 |
| Wilson, 1992 | 34 | 51 | 0.6667 | 0.0346 | 0.5989 | 0.7344 |
| Selesnick, 1993 | 33 | 35 | 0.9429 | 0.0417 | 0.8611 | 1.0246 |
| Dornhoffer, 1994 | 65 | 70 | 0.9286 | 0.0295 | 0.8707 | 0.9864 |
| Gordon, 1995 | 92 | 105 | 0.8762 | 0.0241 | 0.8290 | 0.9234 |
| Chandrasekhar, 1995 | 182 | 197 | 0.9239 | 0.0176 | 0.8894 | 0.9583 |
| Naessens, 1996 | 14 | 15 | 0.9333 | 0.0637 | 0.8084 | 1.0582 |
| Stanton, 1996 | 107 | 111 | 0.9640 | 0.0234 | 0.9180 | 1.0099 |
| Zappia, 1997 | 106 | 111 | 0.9550 | 0.0234 | 0.9090 | 1.0009 |
| Godey, 1998 | 82 | 89 | 0.9213 | 0.0262 | 0.8701 | 0.9726 |
| El-Kashlan, 2000 | 23 | 25 | 0.9200 | 0.0494 | 0.8232 | 1.0168 |
| Haapaniemi, 2000 | 37 | 38 | 0.9737 | 0.0400 | 0.8952 | 1.0522 |
| Magdziarz, 2000 | 149 | 154 | 0.9675 | 0.0199 | 0.9285 | 1.0065 |
| Marangos, 2001 | 211 | 261 | 0.8084 | 0.0153 | 0.7785 | 0.8384 |
| Schmidt, 2001 | 52 | 58 | 0.8966 | 0.0324 | 0.8330 | 0.9601 |
| Lajtman, 2002 | 83 | 86 | 0.9651 | 0.0266 | 0.9129 | 1.0173 |
| Rupa, 2003 | 4 | 4 | 1.0000 | 0.1234 | 0.7581 | 1.2419 |
| Cueva, 2004 | 22 | 31 | 0.7097 | 0.0443 | 0.6228 | 0.7966 |
| Grayeli, 2008 | 644 | 676 | 0.9527 | 0.0095 | 0.9341 | 0.9713 |
| Shih, 2009 | 30 | 30 | 1.0000 | 0.0451 | 0.9117 | 1.0883 |
| Overall Sensitivity | 3098 | 3314 | 0.9348 | 0.0043 | 0.9264 | 0.9432 |
| Study | TN | TN + FP | Specificity | SE (Specificity) | 95% CI Lower | 95% CI Upper |
|---|---|---|---|---|---|---|
| Clemis, 1979 | 80 | 115 | 0.6957 | 0.0358 | 0.6255 | 0.7658 |
| Glasscock, 1979 | 169 | 188 | 0.8989 | 0.0280 | 0.8441 | 0.9538 |
| Terkildsen, 1981 | 164 | 171 | 0.9591 | 0.0294 | 0.9015 | 1.0166 |
| Cashman, 1983 | 61 | 68 | 0.8971 | 0.0466 | 0.8058 | 0.9883 |
| Kotlarz, 1992 | 130 | 167 | 0.7784 | 0.0297 | 0.7202 | 0.8367 |
| Stanton, 1996 | 1137 | 1370 | 0.8299 | 0.0104 | 0.8096 | 0.8503 |
| Rupa, 2003 | 46 | 72 | 0.6389 | 0.0452 | 0.5502 | 0.7276 |
| Cueva, 2004 | 208 | 281 | 0.7402 | 0.0229 | 0.6953 | 0.7851 |
| Overall Specificity | 1995 | 2432 | 0.8203 | 0.0078 | 0.8051 | 0.8356 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
