Article Text
PDF
Abstract
Background The PerfusiOn imaging Selection of Ischemic sTroke patIents for endoVascular thErapy (POSITIVE) trial was designed to evaluate functional outcome in patients with emergent large vessel occlusion (ELVO) presenting within 0–12 hours with pre-specified bifurcated arms of early and late window presentation, who were selected for endovascular thrombectomy with non-vendor specific commercially available perfusion imaging software. Recent trials demonstrating the benefit of thrombectomy up to 16–24 hours following ELVO removed equipoise to randomize late window ELVO patients and therefore the trial was halted.
Methods Up to 200 patients were to be enrolled in this FDA-cleared, prospective, randomized, multicenter international trial to compare thrombectomy and best medical management in patients with ELVO ineligible for or refractory to treatment with IV tissue plasminogen activator (IV-tPA) selected with perfusion imaging and presenting within 0–12 hours of last seen normal. The primary outcome was 90-day clinical outcome as measured by the raw modified Rankin Scale (mRS) with scores 5 and 6 collapsed (mRS shift analysis).
Results The POSITIVE trial suspended enrollment with the release of results from the DAWN trial and was stopped after the release of the DEFUSE 3 trial results. Thirty-three patients were enrolled (21 for medical management and 12 for thrombectomy). Twelve of the 33 patients were enrolled in the 6–12 hour cohort. Despite the early cessation, the primary outcome demonstrated statistically significant superior clinical outcomes for patients treated with thrombectomy (P=0.0060). The overall proportion of patients achieving an mRS score of 0–2 was 75% in the thrombectomy cohort and 43% in the medical management cohort (OR 4.00, 95% CI 0.84 to 19.2).
Conclusion POSITIVE supports the already established practice of delayed thrombectomy for appropriately selected patients presenting within 0–12 hours selected by perfusion imaging from any vendor. The results of the POSITIVE trial are consistent with other thrombectomy trials. The statistically significant effect on functional improvement, despite the small number of patients, reinforces the robust benefits of thrombectomy.
Clinical trial registration NCT01852201
- stroke
- CT perfusion
Data availability statement
Data are available upon reasonable request. Data for this study are available upon reasonable request.
Statistics from Altmetric.com
Introduction
Stroke is the fifth leading cause of mortality in the USA, accounting for approximately 133 000 deaths annually and costing an estimated $40 billion per year in healthcare costs.1 2 Eighty-seven percent of all strokes are ischemic, and those caused by proximal emergent large vessel occlusion (ELVO) are the most devastating, leading to mortality rates as high as ~30% if left untreated.3 Timely recanalization in acute ischemic stroke (AIS) helps restoration of flow to the threatened tissue (penumbra) and is an important determinant of outcome. Intravenous thrombolysis with tissue plasminogen activator (IV-tPA) offers potential therapeutic benefit if administered within 4.5 hours of symptom onset.4–6
Multiple clinical trials have proven the overwhelming benefit of thrombectomy for the treatment of ELVO in patients with AIS.7–14 Furthermore, some recent reports have suggested that, with specific image-based selection (iSchemaView RAPID, Menlo Park, California, USA), patients may be effectively treated 6–24 hours following symptom onset with good outcome.15 16 Since these data became available, additional studies have been performed to evaluate the application of such selection criteria to other populations.17 Additionally, other investigators have provided single-arm data suggesting that alternate selection paradigms may also be of use in late window patients.18 19
The PerfusiOn imaging Selection of Ischemic sTroke patIents for endoVascular thErapy (POSITIVE) trial was an FDA-cleared physician-run trial designed to evaluate the safety and efficacy of thrombectomy to treat AIS due to ELVO 0–12 hours following symptom onset based on physiological imaging selection.
Methods
Study design
POSITIVE (NCT01852201) was a pragmatic multicenter, randomized, prospective phase 3 trial comparing thrombectomy with best medical management (MM) in patients with AIS ineligible for or, in the latter portion of the trial, refractory to treatment with IV alteplase, selected using institutional perfusion imaging protocols and presenting within 0–12 hours from symptom onset. POSITIVE used the latest generation (FDA-approved) thrombectomy devices and each enrolling site obtained Institutional Review Board or ethics committee approval. The trial was designed with anticipation that early window evidence would become available prior to late window treatment and therefore staged halting of the early window was pre-specified should such evidence become available. Following publication of trials demonstrating the benefit of thrombectomy for 0–6 hours, POSITIVE continued as a trial for patients presenting 6–12 hours after onset only.
Patients
Patients aged ≥18 years with National Institutes of Health Stroke Scale (NIHSS) score ≥8 at the time of presentation with neuroimaging demonstrating large vessel proximal occlusion (distal internal carotid artery (ICA) through middle cerebral artery (MCA) M1), a pre-event modified Rankin Scale (mRS) score 0–1, and presenting within 0–12 hours of symptom onset were included. Enrollment also required the presence of a significant salvageable penumbral region via physiologic imaging (CT/CT angiography (CTA)/CT perfusion imaging (CTP) or MRI/MR angiography (MRA)/MR perfusion imaging (MRP)) at presentation, consistent with each center’s institutional protocol. Written informed consent was obtained for all enrolled patients.
Imaging
Imaging with CT or MR was required to exclude acute intracranial hemorrhage. Additional anatomic and physiologic imaging with CT or MRP was performed on patients who did not have evidence of significant ischemia on initial scans, which included significant mass effect with midline shift, large (more than one-third of the MCA) regions of clear hypodensity on the baseline CT scan or Alberta Stroke Program Early CT Score (ASPECTS) of <7. Anatomic imaging used CTA with contrast bolus imaging to visualize the vessels of the head as well as the presence of collateral circulation. Similar anatomic cerebral imaging could be performed with MRA. The studies needed to demonstrate an acute major vessel intracranial anterior circulation occlusion (ICA or MCA) for enrollment.
Physiologic imaging was performed using a contrast bolus tracking technique with either MR or CT technologies. Post processing was in accordance with the institution's routine perfusion evaluation, with deconvolution algorithms preferred. Patient selection was based on the presence of a significant mismatch between the region of infarction depicted on the routine brain imaging scan (extent of CT low attenuation or diffusion-weighted intensity hyperintensity) and/or low cerebral blood volume on perfusion maps as per the institution's standard of care. The viable ischemic penumbra was determined as the region of tissue at risk as determined on the cerebral blood flow and mean transit time (or Tmax) maps minus the region of infarction previously identified. Site-specific perfusion thresholds were chosen to include only significant prolongation in transit times or reduction in blood flow, such that minor perfusion changes reflecting benign oligemia are not included in ‘at risk’ tissue estimates. The qualitative presence of at least two color change differences from the normal was considered significant on the color rainbow scale perfusion maps or, if quantitative measures were used, then an absolute cerebral blood volume <2 mL/100 g reflected the infarct core and relative mean transit time >145% (or absolute value of >6 s) of the contralateral hemisphere most accurately reflected the penumbra. It was recommended that there be at least 50% volume of tissue at risk or an identified eloquent region at risk. Consistent with POSITIVE’s intent to be a pragmatic design, the ultimate 'go' versus 'no-go' decision was intentionally and a priori left to individual practitioners, but they were required to use perfusion assessment to inform their decision. The presence of greater than one-third territory MCA infarction, ASPECTS <7, or acute infarction >35 mL volume (from recent data from the MR RESCUE trial) was considered a boundary for exclusion.
Randomization
Patients who met the inclusion and exclusion criteria and consented to participate were randomly assigned by REDCap, a central web-based system, in a 1:1 manner to treatment with either best MM or combined best MM and thrombectomy with a stratified randomization by NIHSS (<20 vs ≥20) within each center to balance randomization. Data on each patient were collected at the time of enrollment and treatment, and at subsequent follow-up visits (24 hours (±12 hours) post-randomization, 7 days post-randomization or discharge (whichever came first), 30 days (±14 days) post-randomization follow-up, and 90 days (±14 days) post-randomization follow-up).
Intervention details
Once eligibility was determined, patients were randomized to the control (best MM) or the intervention (thrombectomy) group. Best MM consisted of appropriate management of blood sugar levels, blood pressure, and hemodynamic status according to the standard of practice at each institution. For the thrombectomy arm, all FDA-cleared cerebral aspiration and stent retriever thrombectomy devices were allowed. Endovascular intervention was performed under either general anesthesia or conscious sedation based on the treating physician’s standard practice. Thrombectomy was performed with the operator’s standard thrombectomy technique using aspiration or a stent retriever, separately or in combination.
Outcomes
The primary outcome measure was 90-day global disability assessed using raw mRS scores. Secondary outcome measures included 90-day global disability in the 6–12 hour cohort assessed via the overall distribution of mRS, the proportion of patients with good functional recovery for the 6–12 hour cohort at 90 days defined by mRS 0–2, mortality at 30 and 90 days, intracranial hemorrhage (ICH) with neurological deterioration (NIHSS worsening >4) within 24 hours of randomization, procedure-related serious adverse events (including neurological deterioration, angiographic or thrombectomy procedural complications, symptomatic ICH, or malignant or cerebral edema or subarachnoid hemorrhage), and arterial revascularization defined as modified Thrombolysis in Cerebral Infarction (mTICI) 2b or 3 following device use.
Statistical analysis
Statistical power and sample size estimation were conducted using data from previously published clinical trials under the proportional odds ordinal logistic model. The differences between groups were assessed with the Student’s t-test for continuous variables and Fisher’s exact test for categorical variables. For correlations in subanalyses, Pearson’s correlation coefficient was used. Statistical analysis of the primary endpoint was initially pre-specified to be performed with a proportional odds ordinal logistic model. Core laboratory adjudicated data for imaging were used when available. In the case of missing core laboratory imaging or if the reviewer was unable to make a determination, self-reported data were used in conjunction with core laboratory data. Due to the sample size limitations because of early study termination, primary endpoint analysis was also conducted using the Cochran–Mantel–Haenszel (CMH) test using the raw mRS scores for further confirmation of validity. Both the initial pre-specified analysis and the more statistically appropriate CMH analysis are presented. All efficacy and safety outcome measures were analyzed under the intent-to-treat (ITT) (table 1) . The statistical significance was set at a two-sided alpha level of 0.05.
Safety outcomes
Subgroup analysis
The subgroup analysis was performed to compare the two randomized groups in an ITT fashion, evaluated at a two-sided alpha level of 0.01. The ancillary endpoints were performed for clinical efficacy (using improvement in NIHSS of ≥10 points or NIHSS ≤4, NIHSS at discharge/day 7 post-randomization, functional independence determined using mRS, time at which functional independence was achieved, requirement for hemicraniectomy) and technical efficacy.
Results
Study termination
The POSITIVE trial commenced in September 2013 with the aim of assessing the superiority of mechanical thrombectomy over best MM in patients with AIS presenting within 12 hours of symptom onset or time last seen normal. However, since the trial began, evidence of overwhelming benefit of thrombectomy for patients presenting within 6 hours of AIS onset resulted in POSITIVE halting enrollment of patients presenting within the early window in November 2014 and enrolling only in the late time window of 6–12 hours. The trial was subsequently halted altogether after publication of the DAWN and DEFUSE 3 studies in 2018. In total, 21 patients were enrolled in the early window and 12 patients were enrolled in the late window.
Patient characteristics
Overall, 33 patients were enrolled at 11 centers over a period of 5 years between September 2013 and May 2018. Among these, 21 were randomized to maximal MM while 12 were randomized to thrombectomy. The baseline characteristics and demographics of the patient population are shown in table 2. No patients received IV alteplase. All patients assigned to the thrombectomy arm had a thrombectomy procedure performed. One patient assigned to the MM arm was taken for angiography to receive thrombectomy at the physician’s discretion; however, prior to performing the thrombectomy the thrombus was noted to be dissolved and no thrombectomy was performed.
Demographic characteristics
Center imaging characteristics
Enrolling centers were asked to provide the specific perfusion assessment tool used over the course of the trial. All centers reported their standard of care physiologic imaging tools. They reported using the following proprietary perfusion software: 27.3% Siemens; 27.3% RAPID; 18.2% VITAL; 9.1% OLEA; 9.1% GE; and 9.1% used multiphase CTA.20 Core laboratory adjudication was available for all but three patients’ selection imaging. All three of these patients were in the MM arm.
Primary outcomes
Patients with AIS ineligible for or refractory to treatment with IV-tPA with appropriate image selection treated with mechanical thrombectomy within 0–12 hours of symptom onset had less stroke-related disability than those treated with best MM (P=0.0060; OR 10.6, 95% CI 1.97 to 56.90). As an additional robustness check due to the limited sample size because of early study termination, the CMH test was performed (table 3) and confirmed the distribution of functional outcome and global disability assessed via the 90-day mRS score was significantly different for patients with AIS treated with thrombectomy compared with those who received MM alone (P=0.0349). The overall proportion of patients achieving an mRS score of 0–2 was 75% in the thrombectomy cohort and 43% in the MM cohort (OR 4.00, 95% CI 0.84 to 19.2).
Efficacy outcomes
Secondary outcomes
The percentage of patients with good functional recovery in the 6–12 hour cohort defined as mRS 0–2 at 90 days was 50% in the thrombectomy group and 37.5% in the MM group (P=1.0). No patients experienced an ICH with neurological deterioration within 24 hours of randomization. No patients in the MM group experienced a procedure-related serious adverse event compared with one patient in the thrombectomy group (P=0.36).
Safety outcomes
Mortality at 30 days was 12% in the MM group and 0% in the thrombectomy group (P=0.5; table 3). At 90 days, mortality was 19.1% in the MM group and 8.3% in the thrombectomy group (OR 2.6, 95% CI 0.25 to 26.31, P=0.63). At 24 hours post-randomization, no patients had any symptomatic ICH in either group, although 9.5% of patients who received MM and 33% of patients who received thrombectomy experienced asymptomatic ICH (table 3). ICH occurred in 14% of patients in the MM group and in 17% of patients in the thrombectomy group within 90 days of hospital discharge (including discharge) (table 3). The rates of ICH at 24 hours post-randomization and at 90 days of hospital discharge did not differ significantly between the two arms.
Subgroup analyses
Clinical efficacy outcomes
The ability to assess outcomes in subgroup analyses was hindered by the limited sample size as a result of early study termination. For the purpose of subgroup analysis, no adjustments for multiple comparisons were performed. The proportion of patients achieving a NIHSS ≥10 point improvement or an absolute NIHSS ≤4 was greater in the thrombectomy group (92%) than in the MM group (50%) (P=0.0232). The mean maximum documented change in NIHSS was 7.7±7.0 in the MM group and 12±5.7 in the mechanical thrombectomy group (P=0.06). Patients in the MM group had a discharge/7-day post-randomization NIHSS of 9.9±6.5 compared with 5.3±6.9 for the thrombectomy group (P=0.09). No patients in either arm underwent hemicraniectomy (table 4).
Clinical efficacy
Technical outcomes
Of the 12 patients assigned to the thrombectomy arm, 10 (83%) achieved arterial revascularization with an mTICI score of 2b or 3. One patient randomized to the MM group underwent angiography with demonstrated spontaneous arterial revascularization.
Discussion
The results of the POSITIVE trial provide further confirmation that patients with ELVO, ineligible for or refractory to treatment with IV-tPA, with appropriate physiologic image selection treated with thrombectomy within 0–12 hours of symptom onset have less stroke-related disability and improved good functional outcomes compared with those treated with best MM.
The findings of this study are consistent with other thrombectomy studies. Several studies have demonstrated superior functional outcomes in patients undergoing thrombectomy for ELVO. In 2015, five large multicenter prospective randomized clinical trials (MR CLEAN, EXTENDIA, ESCAPE, SWIFT PRIME, and REVASCAT) were published supporting the efficacy and safety of mechanical thrombectomy in the treatment of AIS.8 9 21–23 Subsequent trials further demonstrated the benefit of thrombectomy, cementing the finding that thrombectomy is superior to intravenous thrombolysis alone.13
More recently, the role of thrombectomy in anterior circulation large vessel occlusion beyond the 6 hour window was evaluated. The DAWN (Diffusion Weighted Imaging (DWI) or CT Perfusion (CTP) Assessment with Clinical Mismatch in the Triage of Wake Up and Late Presenting Strokes Undergoing Neurointervention) and DEFUSE 3 (Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke 3) trials have studied the benefit of thrombectomy in anterior circulation large vessel occlusion stroke beyond 6 hours from symptom onset.15 16 Both trials demonstrated that patient selection with the aid of advanced physiologic imaging led to dramatic improvement in clinical outcomes for patients treated with thrombectomy up to 24 hours from stroke onset. These trials firmly established the value of extended window thrombectomy. The outcome of the DAWN trial resulted in the halting of enrollment for the POSITIVE trial. The results of DEFUSE 3 led to the early termination of the POSITIVE trial. Interestingly, DAWN and DEFUSE 3 required the same single-vendor imaging software in their studies, and therefore it remains a debated topic as to whether the benefit observed in these trials was specific to a single perfusion imaging software or is broadly applicable to other perfusion imaging software. Additional work on the 'real world' impact of treating delayed presentation strokes and predictive factors of poor outcomes in the challenging population including physiologic imaging has been well documented.24 25
POSITIVE was a pragmatic trial that allowed centers to use whichever perfusion imaging that was their standard of care. Of the total cohort, 12/33 patients were enrolled in the 6–12 hour cohort and, of those, only 4/12 were randomized to thrombectomy. It is worth noting that nine of the 0-6 hour cohort were actually randomized beyond 6 hours from onset, resulting in a total of 21 patients randomized after six hours from onset. As a consequence, no definitive conclusions can be drawn. The results of the POSITIVE trial raise the possibility that non-standardized perfusion imaging may be an acceptable option to use in delayed window patient selection. If true, then such a finding would potentially be of value because it reflects the reality of how physiology-based selection occurs widely in clinical practice, where it is not restricted to a single vendor’s software. Unfortunately, due to the early termination of POSITIVE and the resultant low number of subjects, these data are extremely limited and such conclusions cannot be drawn from POSITIVE alone. Further work is certainly indicated. For instance, it may also be possible that, if patients were not selected based on perfusion imaging, this trial could still have yielded the same results. It was the intent of POSITIVE to perform detailed evaluation of the perfusion data parameters and potentially elucidate differences between various selection criteria. However, given the heterogeny of imaging methodology and patient presentation coupled with the low number of patients, any meaningful assessment of the detailed perfusion data was not possible. We further acknowledge that the statistically significant result and strong treatment effect is significantly stronger than that observed in the HERMES meta-analysis, and the early termination of the trial may have contributed to the stronger than expected treatment effect. More insight will likely become available when the AURORA delayed window patient-level meta-analysis is completed.
In view of the small number of total patients, POSITIVE may also be subject to selection bias contributing to generalizability. Given that all patients were randomized after enrollment, the benefit observed should not be subject to selection bias; however, the population enrolled may lack generalizability.
Despite the limited sample size as a result of early termination, we still observed a significant benefit with thrombectomy in patients up to 12 hours from last known well. We were also able to ascertain a potential effect of thrombectomy on NIHSS improvement. These results speak to the power of thrombectomy—that with limited enrollment, a pronounced statistically significant effect was observed.
Conclusions
Despite early termination, the POSITIVE trial met its primary endpoint with a limited sample size and demonstrated a statistically significant benefit for thrombectomy. The results of this trial are consistent with other thrombectomy trials. The statistically significant effect on functional improvement, despite the small number of patients, reinforces the robust benefits of thrombectomy.
Data availability statement
Data are available upon reasonable request. Data for this study are available upon reasonable request.
Ethics statements
Patient consent for publication
References
Footnotes
Twitter @AdamArthurMD, @stanfordNRAD
Correction notice This article has been corrected sinec it appeared Online First. Funding statement has updated to "This work was supported by Canon Medical Systems, Medtronic Neurovascular, Penumbra, and Stryker Neurovascular", from "This work was supported by Medtronic".
Contributors All authors made substantial contributions to the study and were involved in the drafting and revising of this manuscript.
Funding This work was supported by Canon Medical Systems, Medtronic Neurovascular, Penumbra, and Stryker Neurovascular.
Competing interests JM is a consultant for Endostream, Viseon, Imperative Care, RIST, Synchron, Viz.ai, Perflow, CVAid, and Cerebrotech, and a shareholder in Rebound, BlinkTBI, Endostream, Viseon, Imperative Care, Serenity, Cardinal Consulting, Q'Apel, NTI, RIST, Synchron, Viz.ai, and Cerebrotech. AHS has two NIH grants as well as financial interest in Adona Medical, Amnis Therapeutics, BlinkTBI, Buffalo Technology Partners, Cardinal Consultants, Cerebrotech Medical Systems, Cognition Medical, Endostream Medical, Imperative Care, International Medical Distribution Partners, Neurovascular Diagnostics, Q’Apel Medical, Radical Catheter Technologies, Rebound Therapeutics Corp, Rist Neurovascular, Sense Diagnostics, Serenity Medical, Silk Road Medical, Spinnaker Medical, StimMed, Synchron, Three Rivers Medical, Vastrax, ICIS, and Viseon; consultant for Amnis Therapeutics, Boston Scientific, Canon Medical Systems USA, Cerebrotech Medical Systems, Cerenovus, Corindus, Endostream Medical, Imperative Care, Integra, Medtronic, MicroVention, Minnetronix Neuro, Northwest University – DSMB Chair for HEAT Trial, Penumbra, Q’Apel Medical, Rapid Medical, Rebound Therapeutics Corp, Serenity Medical, Silk Road Medical, StimMed, Stryker, Three Rivers Medical, VasSol, and W L Gore & Associates; PI or on the Steering Committee for Cerenovus LARGE Trial and ARISE II Trial, Medtronic SWIFT PRIME and SWIFT DIRECT Trials, MicroVention FRED Trial and CONFIDENCE Study, MUSC POSITIVE Trial, Penumbra 3D Separator Trial, COMPASS Trial, and the INVEST Trial. DF has consulting agreements with Medtronic, Cerenovous, Microvention, Stryker, Balt USA, Marblehead, RAPID.AI, and RAPID Medical; research support from Cerenvous, Microvention, Penumbra, Stryker, Balt USA, and Siemens; stock holder in Vascular Simulations, Neurogami, and Marble Head; board member at Vascular Simulations; honorarium at Qapel Medical. MJA is a consultant for Stryker Neurovascular, Penumbra, and Device Proctor for Medtronic. ASA is a consultant for Balt, Johnson and Johnson, Medtronic, Microvention, Penumbra, Scientia, Siemens, Stryker; research support from Balt, Medtronic, Microvention, Penumbra, and Siemens; shareholder in Azimuth, Bendit, Cerebrotech, Endostream, Magneto, Marblehead, Neurogami, Serenity, Synchron, Triad Medical, and Vascular Simulations. BWB is a consultant for Penumbra, Stryker, Medtronic, Cerenovus, Route 92, Artio Medical, 880 Medical, and Rapid Medical; shareholder in Penumbra (stock divested), Route 92, Artio Medical, Rapid Medical, and Marblehead. MTF has grant funding from NIH, Medtronic, Microvention, Stryker, Penumbra, EndoPhys, and Genentech; serves as a consultant for Medtronic, Stryker, Balt USA, VIZ.AI, Genentech, Cerenovus, Corindus, and EndoPhys; consultant for Medtronic, Stryker, MicroVention, Cerenovous, Balt, Phenox, SAB at eLum, Three Rivers Medical, Shape Medical, MiVI; stock holder in RIsT, Inneuro Co, Deinde, eLum, Cerebrotech, Endostream, Three Rivers Medical, BendIT, Blink TBI, Serenity, Scientia. JL has consulting agreements with Penumbra and Stryker. ASP has research grants from Medtronic, Cerenovus, and Stryker; consultant for Microvention, CereVasc, Qpel, Scientica, Arsenal Medical, Merit; stockholder in NTI, Galaxy, and InNeuroCo. ATR has consulting agreements with Stryker Neurovascular, Cerenovus and Microvention. KW is a consultant for Medtronic, Microvention, Penumbra, and Cerenovus. AT is a consultant for Cardinal Consulting, Cerebrotech, Cerenovus, Corindus Robotics, Endostream Medical, Medtronic, Siemens, Imperative Care, Three Rivers Medical, Vastrax, Shape Memory, Serenity Medical, 880 Medical, and Q’Apel; stock holder in Cerebrotech, Endostream Medical, Imperative Care, Three Rivers Medical, Vastrax, Pipe Therapeutics, Q’Apel, Shape Memory, Synchron, Serenity Medical, Blink TBI, Echovate, RIST, Apama, VIZ AI, Early Bird Medicak, Radical Medical, Spinnaker Medical; co-founder of PipeTherapeutics, Neuro Technology Investors (NTI), National Education and Research Center (NEAR), Imperative Care; on the board of BlinkTBI and provides expert testimony for Corindus Vascular Robotics.
Provenance and peer review Not commissioned; externally peer reviewed.