Investigating heterogeneity of treatment effect for convalescent plasma in severe covid-19: secondary analysis oe a randomized controlled trial

SESSION TITLE: Acute COVID-19 and Beyond: from Hospital to HomeboundSESSION TYPE: Original InvestigationsPRESENTED ON: 10/18/2022 02:45 pm - 03:45 pmPURPOSE: Minimally-biased clustering (MBC) has identified hypoinflammatory (hypo-I) and hyperinflammatory (hyper-I) subphenotypes in ARDS. The hyper-I type exhibits higher inflammatory markers, clinical severity, and mortality. Similar subphenotypes were recently identified in COVID-19-related ARDS. Lower PCR cycle threshold was associated with higher mortality in the hypo-I type, implying an association between viral load (VL) and clinical outcomes in patients with dampened inflammatory responses. In a recent randomized clinical trial (RCT), convalescent plasma (CP) improved survival in severe COVID-19. We hypothesized that the anti-viral effect of CP would more significantly benefit patients without acute hyperinflammation, for whom VL may be associated with mortality.METHODS: From 4/2020-11/2020, 223 adults >18 years of age in New York and Rio de Janeiro with laboratory-confirmed severe COVID-19 were enrolled in a double-blind RCT evaluating the efficacy of CP. 150 patients received CP; 73 received control plasma. Hierarchical clustering (HC) of clinical and laboratory data was used to identify sub-groups in the study population. Primary and secondary outcomes were clinical status at 28 days by modified WHO ordinal score (higher scores indicating worse status) and 28-day mortality. Welch’s t-tests, chi-squared tests, and Fisher’s exact tests were used to compare clinical and laboratory data across clusters. Proportional odds and logistic regression were used to assess the association between cluster-derived subgroups and outcome and the interaction between subgroups and randomized treatment assignment.RESULTS: HC identified two clusters (C1; N=156 and C2; N=67) in the population. Patients in C2 had significantly higher markers of inflammation (sedimentation rate, C-reactive protein, interleukin-6), coagulation (D-dimer), and cardiac injury (cardiac troponin) as well as relative lymphopenia, hypoalbuminemia, and lower bicarbonate. At 28 days, patients in C2 had significantly worse clinical status (OR of 1-pt ordinal score increase 3.10, 95% CI 1.72-5.60, p=0.0002) and higher mortality (28.4% vs. 11.5%, OR 3.03, 95% CI 1.47-6.26, p=0.003). There was no significant between-cluster heterogeneity of CP treatment effect on either ordinal score (OR 0.56, 95% CI 0.16-1.95, p=0.36) or mortality (OR 0.52, 95% CI 0.12-2.30, p=0.38).CONCLUSIONS: C2 exhibited elevated inflammatory markers and lymphopenia indicative of an acute hyperinflammatory response. C2 exhibited poorer clinical status and higher mortality at 28 days. There was no evidence of significant heterogeneity of CP treatment effect on 28-day clinical outcomes.CLINICAL IMPLICATIONS: The previously shown mortality benefit of CP in severe COVID may not differ based on inflammatory state. Using MBC methods on larger samples, e.g., patient data from a meta-analysis of CP trials, may reveal a significant impact of inflammatory state on CP effect.DISCLOSURES: No relevant relationships by Matthew CummingsReceived a grant sub-award from Amazon relationship with Amazon Please note: 4/2020 -12/2020 Added 03/10/2022 by Max O'Donnell, value=Grant/Research SupportNo relevant relationships by Tejus SatishNo relevant relationships by Allison Wolf SESSION TITLE: Acute COVID-19 and Beyond: from Hospital to Homebound SESSION TYPE: Original Investigations PRESENTED ON: 10/18/2022 02:45 pm - 03:45 pm PURPOSE: Minimally-biased clustering (MBC) has identified hypoinflammatory (hypo-I) and hyperinflammatory (hyper-I) subphenotypes in ARDS. The hyper-I type exhibits higher inflammatory markers, clinical severity, and mortality. Similar subphenotypes were recently identified in COVID-19-related ARDS. Lower PCR cycle threshold was associated with higher mortality in the hypo-I type, implying an association between viral load (VL) and clinical outcomes in patients with dampened inflammatory responses. In a recent randomized clinical trial (RCT), convalescent plasma (CP) improved survival in severe COVID-19. We hypothesized that the anti-viral effect of CP would more significantly benefit patients without acute hyperinflammation, for whom VL may be associated with mortality. METHODS: From 4/2020-11/2020, 223 adults >18 years of age in New York and Rio de Janeiro with laboratory-confirmed severe COVID-19 were enrolled in a double-blind RCT evaluating the efficacy of CP. 150 patients received CP; 73 received control plasma. Hierarchical clustering (HC) of clinical and laboratory data was used to identify sub-groups in the study population. Primary and secondary outcomes were clinical status at 28 days by modified WHO ordinal score (higher scores indicating worse status) and 28-day mortality. Welch’s t-tests, chi-squared tests, and Fisher’s exact tests were used to compare clinical and laboratory data across clusters. Proportional odds and logistic regression were used to assess the association between cluster-derived subgroups and outcome and the interaction between subgroups and randomized treatment assignment. RESULTS: HC identified two clusters (C1; N=156 and C2; N=67) in the population. Patients in C2 had significantly higher markers of inflammation (sedimentation rate, C-reactive protein, interleukin-6), coagulation (D-dimer), and cardiac injury (cardiac troponin) as well as relative lymphopenia, hypoalbuminemia, and lower bicarbonate. At 28 days, patients in C2 had significantly worse clinical status (OR of 1-pt ordinal score increase 3.10, 95% CI 1.72-5.60, p=0.0002) and higher mortality (28.4% vs. 11.5%, OR 3.03, 95% CI 1.47-6.26, p=0.003). There was no significant between-cluster heterogeneity of CP treatment effect on either ordinal score (OR 0.56, 95% CI 0.16-1.95, p=0.36) or mortality (OR 0.52, 95% CI 0.12-2.30, p=0.38). CONCLUSIONS: C2 exhibited elevated inflammatory markers and lymphopenia indicative of an acute hyperinflammatory response. C2 exhibited poorer clinical status and higher mortality at 28 days. There was no evidence of significant heterogeneity of CP treatment effect on 28-day clinical outcomes. CLINICAL IMPLICATIONS: The previously shown mortality benefit of CP in severe COVID may not differ based on inflammatory state. Using MBC methods on larger samples, e.g., patient data from a meta-analysis of CP trials, may reveal a significant impact of inflammatory state on CP effect. DISCLOSURES: No relevant relationships by Matthew Cummings Received a grant sub-award from Amazon relationship with Amazon Please note: 4/2020 -12/2020 Added 03/10/2022 by Max O'Donnell, value=Grant/Research Support No relevant relationships by Tejus Satish No relevant relationships by Allison Wolf