Schmidt et al (2014) in Germany designed a study to examine confounding factors that affect Ntms performance (Schmidt, 2014). In a 3-part design, investigators differentiated variance due to physiological factors (eg, tissue conductivity, brain rhythms, cognitive state, peripheral sensory input, preinnervation, brain dysfunction) from physical variation of the nTMS device (ie, coil location, orientation, and tilt, stimulation strength). Twenty healthy volunteers participated in 2 experiments to compare targeted stimulation (optimal stimulus location, orientation, and tilt parameters) with nontarget-controlled stimulation. Four healthy volunteers participated in a third experiment of maximal physiological confounding variance (eg, patients were instructed to maximally contract the target muscle). Spatial resolution of nTMS (defined as variation in the area of cortical stimulation that leads to maximum muscle contraction) was found to be approximately 5 mm so that “even small physical fluctuations can confound the statistical comparison of corticospinal excitability measurements” (Schmidt, 2014).  The authors recommended step-wise regression to partition physical from physiological variance in nTMS results and to produce more interpretable data.

 

Most studies of nTMS are small case series of patients with brain tumors (Mangarviti, 2013; Opitz, 2014; Rizzo, 2014), cavernous angiomas (Paiva, 2013),  arteriovenous malformations (Kato, 2014) or other brain lesions; these are not ideal studies to ascertain diagnostic characteristics. Because of the use of nTMS and/or other methods to identify motor or language centers in the cortex to determine surgical approach, the reference standard of direct cortical stimulation (DCS) may be biased; that is, the DCS procedure may be limited or altered because of the tumor resection or other surgical factors. It is not possible to verify all nTMS sites identified, because the surgical field is limited. Because of this necessarily limited verification, it is difficult to ascertain diagnostic characteristics of nTMS. nTMS is being studied as a technique to augment preoperative detection of motor corticospinal tracts (CSTs), which are currently identified using diffusion tensor imaging (DTI), an MRI technique (Conyi, 2014; Frey, 2014). Conti et al (2014) compared the size and location of (cortical) motor maps determined by the cortical end of CSTs, identified using DTI only and nTMS-DTI, to nTMS maps (Conti, 2014). Twenty patients who underwent brain surgery at a single center in Italy were prospectively enrolled. All brain lesions (70% brain tumors [glioma, astrocytoma, glioblastoma multiforme], 20% cavernous angioma, 10% metastasis) were located within 10 mm of the motor cortex. nTMS-DTI was performed the day before surgery, and standard DTI was obtained after surgery using preoperative imaging data. Direct subcortical stimulation (functional tractography) was applied to confirm tract location. Overlap between nTMS cortical maps and cortical end-regions of CSTs was greater with nTMS-DTI compared with standard DTI (90% vs 58%). Direct subcortical stimulation confirmed CST location in all patients. A potential limitation of the study is lack of DCS to confirm nTMS-determined motor maps. Larger comparative studies with clinical outcomes are needed to assess the clinical relevance of these results.

 

A research group in Germany published 2 studies of nTMS for mapping cortical language sites, one in healthy volunteers and one in patients with brain tumors (Krieg, 2014). In a case series of 10 healthy volunteers, nTMS test-retest reliability varied across error type (eg, neologism, semantic error) and cortical region (ie, anterior or posterior), but overall, both intra- and interobserver reliability were low (range of concordance correlation coefficients: intraobserver, –0.222-0.505; interobserver, –0.135-0.588) (Sollmann, 2013). In a case report of 3 patients with language-eloquent brain tumors who underwent nTMS and DCS for both initial surgery and repeat surgery for recurrence, nTMS performance characteristics varied by definition of a positive nTMS finding (ie, a language error made in response to stimulation) (Krieg, 2014).  For positivity defined by error rates (percentage of stimulations that produced errors) ranging from 5% to 25%, sensitivity was 90% to 10%, specificity was 28% to 89%, PPV was 21% to 17%, and NPV was 93% to 82%. Plasticity of language areas in both healthy volunteers and in patients with brain lesions was identified as a source of variation in nTMS studies across time. As noted in one review, the language network appears to spread over both hemispheres, increasing the complexity of presurgical language mapping (Picht, 2014).

 

 

The Nexstim website21 lists 2 single-center studies from Germany that compared clinical outcomes in patients with motor-eloquent brain tumors who underwent surgical resection with or without preoperative nTMS (Frey, 2014; Kreig, 2014). Both studies used historical controls. Frey et al (2014) enrolled 250 consecutive patients who underwent nTMS preoperative mapping and identified 115 similar historical controls (Frey, 2014). Fifty-one percent of the nTMS group and 48% of controls had WHO grade II to IV gliomas; remaining patients had brain metastases from other primary cancers or other lesions. Intraoperative motor cortical stimulation to confirm nTMS findings was performed in 66% of the nTMS group. British Medical Research Council and Karnofsky scales were used to assess muscle strength and performance status, respectively. Outcomes were assessed at postoperative day 7 and then at 3-month intervals. At 3 months follow-up, 6.1% of the nTMS group and 8.5% of controls had new postoperative motor deficits (chi-square test, p=NS); changes in performance status postoperatively also were similar between groups. Other outcomes were reported for patients with glioma only (n=128 nTMS patients, n=55 controls). Based on postoperative MRI, gross total resection was achieved in 59% of nTMS patients and in 42% of controls (chi-square test, p<0.05). At mean follow-up of 22 months (range, 6-62) in the nTMS group and 25 months (range, 9-57) in controls, mean PFS was similar between groups (mean PFS, 15.5 months [range, 3-51] nTMS vs 12.4 months [range, 3-38] controls; statistical test for survival outcomes not specified, p=NS). In the subgroup of patients with low-grade (grade II) glioma (n=38 nTMS patients, n=18 controls), mean PFS was longer in the nTMS group (mean PFS, 22.4 months [range, 11-50] nTMS vs 15.4 months [range, 6-42] controls; p<0.05), and new postoperative motor deficits were similar (7.5% vs 9.5%, respectively; chi-square test, p=NS). Overall survival did not differ statistically between treatment groups. Interpretation of these findings is limited by: the single-center setting (because nTMS is an operator-dependent technology, applicability may be limited), use of historical controls (surgeon technique and practice likely improved over time), selective outcome reporting (survival outcomes in glioma patients only), and uncertain validity of statistical analyses (primary outcome not identified, no correction for multiple testing, statistical tests not identified). In an accompanying editorial, Jensen outlined the following additional issues complicating interpretation of the study results (Jensen, 2014):

 

 

In the second study from Germany, Krieg et al (2014) enrolled 100 consecutive patients who underwent nTMS preoperative mapping and identified 100 historical controls who were matched for tumor location, preoperative paresis, and histology (Krieg, 2014). Most patients had glioblastoma (37%), brain metastasis (24%), or astrocytoma (29%). Data analysis was performed blinded to group assignment. The primary efficacy outcome was not specified. Median follow-up was 7.1 months (range, 0.2–27.2) in the nTMS group and 6.2 months (range, 0.1–79.4) in controls. Incidence of residual tumor by postoperative MRI was less in the nTMS group compared with controls (22% vs 42%; odds ratio [OR]=0.38 [95% CI, 0.21 to 0.71]). Incidence of new surgery-related transient or permanent paresis did not differ between groups. However, “when also including neurological improvement [undefined] in the analysis,” more patients in the nTMS group improved (12% nTMS vs 1% controls), and similar proportions of patients worsened (13% nTMS vs 18% controls) or remained unchanged (75% nTMS vs 81% controls; Mann-Whitney-Wilcoxon test, p=0.006). Limitations of this study include the single-center setting, use of historical control, uncertain outcome assessments (“neurological improvement” not defined), and uncertain validity of statistical analyses (primary outcome not identified, no correction for multiple testing).

 

In his editorial, Jensen challenged the assertion that randomized trials of nTMS would be unethical, suggesting instead that equipoise exists about the best among several noninvasive mapping techniques (fMRI, magnetoencephalography, nTMS) (Jensen, 2014). Krieg et al concurred, stating that “a randomized trial on the comparison with the gold standard of intraoperative mapping seems mandatory to gain level I evidence for this modality” (Krieg, 2014).

 

 

 

Tarapore and colleagues evaluated the safety of nTMS in a large multicenter series of 733 patients (Tarapore, 2016). Patients had tumors in eloquent or perieloquent regions of the brain and underwent nTMS as part of presurgical planning. nTMS frequencies of 5, 7 and/or 10 Hz were used. A total of 537 patients underwent single pulse motor mapping, 38 had repetitive-pulse language mapping, and 158 had both of these. nTMS was successfully completed in all patients. No seizures (focal, complex or generalized) were reported and no patients reported hearing changes, cognitive or neuropsychological changes, or other transient adverse effects. Headache, reported by 28 patients (6%). was the most commonly reported adverse effect. A total of 141 of 196 patients (72%) completed questionnaires after the procedure and 131 of these (93%) reported discomfort during the procedure. Using a visual analogue scale (VAS) of 1 to 10, 33 of 131 (25%) patients reported a VAS of 1-3 and the remaining 98 (75%) reported a VAS >3.

 

 

A second study by this research group, with some overlap in enrolled patients, was published by Krieg and colleagues (Krieg, 2015).  This study prospectively enrolled 70 patients who underwent nTMS and matched them with a historical control group of 70 patients who did not have preoperative nTMS. All patients had motor eloquently located supratentorial high-grade gliomas (HGG) and they all underwent craniotomy in the single department by the same group of surgeons. As in the 2014 study by Krieg and colleagues, patients were matched by tumor location, preoperative paresis and histology, and the primary outcome was not specified. Outcome assessment was blinded. Craniotomy size was 25.3 cm2 (SD: 9.7cm) in the nTMS group and 30.8 cm2 (SD: 13.2) in the non-nTMS group; the difference in size was statistically significant, p=0.006. There was not a statistically significant difference between groups in the rate of surgery-related paresis, rate of surgery-related complications on MRI or the degree of motor impairment during follow-up. Median overall survival was 15.7 months (SD: 10.9) in the nTMS group and 11.9 months (SD: 10.3) in the non-nTMS group which was not significantly different between groups (p=0.131). Mean survival at 3, 6 and, 9 months was significantly higher in the nTMS group compared with the non-nTMS group and mean survival at 12 months did not differ significantly between groups.

 

One study used concurrent controls, but did not randomize patients to treatment group. Sollman and colleagues matched 25 prospectively enrolled patients who underwent preoperative nTMS but whose results were not available to the surgeon during the operation (group 1) to 25 patients who underwent preoperative nTMS and results were available to the surgeon (group 2) (Sollman, 2015). All patients had language eloquently located brain lesions within the left hemisphere. Primary outcomes were not specified. Three months after surgery, 21 patients in group 1 had no or mild language impairment and 4 patients had moderate to severe language deficits. In group 2, 23 patients had no or mild language impairment and 2 patients had moderate to severe deficits. The difference between groups in post-operative language deficits was statistically significant (p=0.0153). Other outcomes, including duration of surgery, post-operative scores on the Karnofsky performance status scale, percent residual tumor, and peri- and postoperative complication rates did not differ significantly between groups.

 

 

 

 

 

 

 

Annual policy review completed with a literature search using the MEDLINE database through October 2020. No new literature was identified that would prompt a change in the coverage statement.