Robert W. Regenhardt, MD, PhD
Choi K-H, Kim J-H, Kang K-W, Kim J-T, Choi S-M, Lee S-H, et al. HbA1c (Glycated Hemoglobin) Levels and Clinical Outcome Post-Mechanical Thrombectomy in Patients With Large Vessel Occlusion. Stroke. 2018
Mechanical thrombectomy has revolutionized the acute care of patients suffering from stroke secondary to large vessel occlusions. With its proven efficacy up to 24 hours after stroke onset, current research endeavors include broadening the inclusion criteria for this powerful therapy, preventing complications such as reperfusion injury, and increasing favorable outcomes after successful recanalization. Admission hyperglycemia is a known predictor of poor outcomes and increased mortality in patients suffering from acute stroke, including those treated with intravenous thrombolysis and mechanical thrombectomy.
This recent manuscript by Choi et al. published in Stroke analyzed relationships between hemoglobin A1C (HbA1C), a marker of average blood glucose during the preceding 3 months, and clinical outcomes after thrombectomy. The authors examined HbA1C in 534 patients with acute ischemic stroke from large vessel occlusions that underwent thrombectomy in a prospective cohort study. The primary outcome was disability (mRS) at 3 months. Secondary outcomes included 3-month mortality. In addition, several other outcomes were examined within the first 7 days: intracranial hemorrhage, early neurologic deterioration (worsening by ≥2 NIHSS points), and early clinical improvement (by ≥4 NIHSS points).
Dichotomizing their dataset by HbA1C>6.5 (29%) vs not, they showed that patients with elevated HbA1C were less likely to have mRS 0-2 at 3 months (29% vs 42%, P=0.006). Furthermore, Cochrane-Mantel-Haenszel testing for shift analysis showed a significant shift in mRS at 3 months (P=0.005). Mortality at 3 months was also higher in the group with HbA1C>6.5 (21% vs 8%, P<0.001). In addition, significant differences were observed in the first 7 days, including worsening by ≥2 NIHSS points (25% vs 15%, P=0.007) and improving by ≥4 NIHSS points (40% vs 55%, P=0.003). Symptomatic intracranial hemorrhage was not different (7% vs 5%, P=0.316), though there was a trend for more hemorrhagic transformation overall (40% vs 33%, P=0.074).
The magnitude of the differences the authors describe is remarkable. The absolute risk reduction of having A1C<6.5 to prevent mRS 3-6 at 3 months is 71.2%-57.9%=13.3% for a number needed to treat of 7.5. If you consider the magnitude of the effect of thrombectomy itself from the HERMES meta-analysis of the 5 thrombectomy trials of 2015, where the absolute risk reduction to prevent the same level of disability is 73.5%-54.9%=19.5% for a number needed to treat of 5.1, the potential importance of controlling diabetes is astonishingly similar.
While the authors adjusted for confounders (age, sex, hypertension, dyslipidemia, smoking, atrial fibrillation, history of stroke/TIA, initial NIHSS, and recanalization status) in their multivariate analysis, Table 1 shows that there were several noteworthy differences comparing those with HbA1C>6.5 vs not. There was a trend for more hypertension (65% vs 56%, P=0.080), higher triglycerides (98 vs 86, P=0.037), lower HDL cholesterol (44 vs 49, P<0.001), and higher NIHSS at presentation (12.4 vs 11.4, P=0.027). The authors state that this may suggest this group had more metabolic syndrome, which is plausible and perhaps not surprising that these risk factors would be concomitant.
In addition, perhaps as expected, those with HbA1C>6.5 also had higher admission serum glucose (172 vs 128, P<0.001) and higher fasting blood glucose on day 2 (155 vs 120, P<0.001). The authors further analyzed these glucose parameters and showed that both admission and day 2 fasting blood glucose were negatively associated with mRS 0-2 and positively associated with mortality.
In their subgroup analyses, the authors found no heterogeneity in the effect of HbA1C on 3-month mRS according to age, sex, baseline NIHSS, administration of thrombolysis, or location of LVO. However, there was a significant interaction according to recanalization status (P=0.035). The subgroup that was not recanalized to TICI 2b-3 (n=140) did not appear to benefit (OR 1.16, 95% CI 0.44-3.03), while the group that did recanalize (n=394) may be driving the observed effect (OR 0.36, 95% CI 0.22-0.60).
While other studies have shown hyperglycemia is associated with poor outcomes after stroke, this study is the first to also examine HbA1C and recanalization status after thrombectomy. Perhaps those that do not recanalize will have poor outcomes regardless given “completion” of large infarcts. Those that do recanalize may have outcomes dependent on a number of variables. As the authors describe, hyperglycemia may be one such variable and could be linked to an exacerbation of ischemic brain injury, through inflammatory responses, free radical formation, endothelial dysfunction, blood-brain barrier degradation, impaired autoregulation, and reperfusion injury. Furthermore, the association of HbA1C (in addition to admission and day 2 fasting blood glucose) with poor outcomes suggests that there may be chronic pre-stroke injury and biological changes to both the microvasculature and brain parenchyma that make these patients more susceptible to worse outcomes after stroke and impair recovery. Indeed, previous studies have shown that while hyperglycemia during the acute phase is associated with poor outcomes, intensive glycemic control during this acute period did not improve functional outcomes or mortality. The authors state their results suggest there is an urgent need to improve glucose control both before and after stroke.
There are several limitations to the study. Perhaps the most significant is the unknown effects of other comorbidities that often accompany diabetes, such as diet and activity level, although the authors tried to control for some in their multivariate model. Furthermore, the authors remind the reader that a cause-effect relationship cannot be assumed based on these results. There was no inclusion in their model of antidiabetic treatments before or after stroke or the duration of diabetes before stroke (about half of the group with HbA1C>6.5 were unaware they had diabetes). Other limitations inherent to a single center retrospective analysis of prospectively registered cohort data, including lack of randomization, should be considered when interpreting and generalizing these data. Future studies are needed to verify these results and control for other variables that may contribute to poor outcomes in diabetic patients who suffer from stroke.