By Baxter B. Allen, MD
Synopsis: Secondary neurologic decline after large middle cerebral artery (MCA) stroke is common and is associated with worse outcomes. This study aimed to evaluate the utility of quantitative pupillometry to predict neurologic decline before it occurs. Although no model was found to predict decline, stable Neurological Pupil index (NPi) and dilation velocities may provide some reassurance that an imminent decline is unlikely.
Source: Du Y, Pohlmann JE, Chatzidakis S, et al. Quantitative pupillometry predicts neurologic deterioration in patients with large middle cerebral artery strokes. Ann Neurol. 2025;97(5):930-941.
Despite recent advances in care, stroke remains the fifth leading cause of death and the first leading cause of disability in the United States. Ischemic stroke-related mortality has declined over the last two decades but still is associated with an approximately 10% 30-day and 20% one-year mortality rate. A significant cause of mortality is the development of malignant edema, which occurs in up to 30% of large middle cerebral artery (MCA) strokes. Although interventions such as decompressive hemicraniectomy can reduce mortality significantly in these patients (from 80% to as low as 20%), there still is a strong association with severe disability. These patients require close neurologic monitoring to evaluate for secondary neurologic deterioration (ND), which is most associated with worsening cerebral edema.
It has long been known that acute changes in pupillary function are associated with catastrophic events because of elevated intracranial pressure and worsening mass effect. Although bedside examination with a penlight to evaluate the pupillary light reflex has long been the standard, quantitative pupillometry has become increasingly common and allows for more precise evaluation of pupillary function in a less subjective manner. Quantitative pupillometry using a pupillometer will automatically measure pupil size at rest and after light, percent and absolute change in pupil size, constriction and dilation velocity, and latency of pupillary response. The Neurological Pupil index (NPi) is a numeric score ranging from 0 (non-reactive) to 5 (normal) calculated using these data points. It has been shown to be useful in predicting post-discharge outcomes related to multiple acute brain injuries, including ischemia; however, historically, these observations occur after deterioration.
The study by Du et al aimed to evaluate the utility of quantitative pupillometry to predict ND in patients with large MCA strokes before clinically apparent deterioration has occurred. This was a single-center, prospective observational study of adult patients with acute ischemic MCA strokes admitted over a five-year period to the neuroscience intensive care unit at Boston Medical Center. Of 777 patients screened, 71 were found who met the inclusion criteria: stroke greater than one-half the MCA territory, first pupil measurement less than 24 hours from time last known well, first pupil measurement prior to development of midline shift ≥ 5 mm, decompressive hemicraniectomy, or ND.
Exclusion criteria included significant focal encephalomalacia of one-half or more of any vascular territory; other acute stroke, hemorrhage, or brain tumor; or withdrawal of life-sustaining therapies (WLST) within 24 hours of admission without an interval computed tomography (CT) head.
The primary outcome was time-to-ND from last seen well, which was defined as a secondary clinical decline persisting at least six hours or resulting in surgical decompression, with a decreased Glasgow Coma Scale (GCS) score of 2 or more, an increase in National Institutes of Health Stroke Scale (NIHSS) score of 2 or more, an increase of 1 or more in NIHSS motor or consciousness components, or a new non-pharmacologic absent pupillary reflex. Although WLST patients were included, death occurring as the result of WLST was not classified as a neurologic deterioration.
The study focused on finding an optimal threshold of pupillometric change within a 12-hour period prior to clinical ND. Of the 71 patients in the study, 32 experienced ND with a mean onset time of 46.3 hours after last known well time (LKWT). Deterioration was noted to include 29 patients with a GCS decrease ≥ 2, and three patients with an increase in NIHSS motor or consciousness components. Within the study group, 26.8% of patients received a subsequent decompressive hemicraniectomy, and 31% died. ND was mostly the result of worsening cerebral edema (26 patients), with smaller contributions caused by acute respiratory failure (four patients) or hemorrhagic conversion without mass effect (two patients).
Overall, patients with a lower NPi had a significantly higher risk of developing secondary ND. Although a decrease in NPi generally was noted over a 12-hour period prior to deterioration, these changes became more pronounced at the four-hour mark. Most aggregate pupillometry characteristics did not differ between the patients with and without ND, except for dilation velocity (0.38 mm/s in patients with ND; 0.74 mm/s in patients without ND) and constriction velocity (0.89 mm/s in patients with ND; 1.52 mm/s in patients without ND). No model of NPi, changes in NPi, or its related values had a high positive predictive value; however, negative predictive value was high across all metrics, suggesting patients with an NPi > 4, minimal change in NPi over 12 hours, and dilation velocity > 0.49 mm/s are likely at low risk of impending deterioration.
Commentary
The pupillary examination remains a cornerstone for evaluating patients with altered mentation in the critical illness setting. The development of pupillometers has allowed for a more easily quantifiable pupillary exam, and provides additional data beyond an approximate size, “briskly” vs. “sluggishly” reactive, and “equal” or “unequal” with the traditional bedside examination. Additionally, automated pupillometry allows for significantly improved inter- and intrarater reliability, and several reliable variables that can be tracked over time during a patient’s intensive care unit stay.
This study highlights the need for further research into the ability of these variables to predict neurologic changes, which one hopes will allow physicians to proactively intervene to prevent neurologic decline rather than act after the fact to try and reverse a new injury. Although an exact threshold of changes to predict ND is not clearly defined in this small observational study, the high negative predictive value of stable NPis and dilation velocities likely can provide some reassurance that an imminent decline is unlikely.
Baxter B. Allen, MD, is Assistant Professor of Clinical Neurology, Weill Cornell Medicine.
