Quickstroop predicts time to development of overt hepatic encephalopathy and related hospitalizations in patients with cirrhosis

Cirrhosis-related neurocognitive impairment due to covert or minimal hepatic encephalopathy (CHE) affects psychosocial function, increases risk of overt HE (OHE) development, and worsens survival^{1, 2}. However, detection in clinical practice is challenging². One modality used for screening and prediction of outcomes related to cirrhosis is the EncephalApp Stroop, but it can require up to 10 minutes. Furthermore, the assessment comprises of distinct stages of difficulty, with an easier “Off” stage and a more challenging “On” stage³. To alleviate these concerns, QuickStroop, which takes <1 minute, was developed. This uses only the first 2 runs of the Off stage of the EncephalApp Stroop, where # signs presented in red, green, or blue need to be matched quickly to their respective colors⁴. A prior study showed these versions were comparable cross-sectionally to diagnose CHE⁴. However, the utility of Quickstroop to predict cirrhosis-related outcomes is unclear^5–7. Our aim was to determine the ability of QuickStroop to determine time to development of OHE and OHE-related hospitalizations, all-cause hospitalizations, and death in outpatients with cirrhosis.

After IRB approval and consent, we enrolled outpatients with cirrhosis above 18 years of age from two tertiary care centers. We excluded those who did not provide consent, had mini-mental status exam<25, were actively abusing alcohol/illicit drugs, or were color blind. Cirrhosis was diagnosed either by liver biopsy, radiologic evidence or endoscopic evidence of varices or signs of frank decompensation such as ascites, prior OHE, or prior variceal bleeding. We included those with prior OHE, provided that they demonstrated adherence to lactulose or rifaximin for >2 months and were able to consent. QuickStroop was administered in clinic using a standard iPad. CHE on Quickstroop was diagnosed based on age, gender, and education-adjusted norms. We recorded demographics, cirrhosis severity and complications, and liver-related medications.

Time from testing to the development of OHE (defined as grade ≥2 needing treatment initiation or change), OHE-related non-elective hospitalizations (all hospitalizations where OHE was the main reason for admission), all-cause hospitalizations, and death were recorded using chart review and patient interviews⁸. Cox proportional hazard models dividing patients by CHE on QuickStroop for OHE and related hospitalizations were created, considering both unadjusted and adjusted for variables including cirrhosis severity, demographics, medications, and prior complications. Competing risk analysis using the Fine and Gray method to adjust for LT and death was also performed.

We enrolled 250 patients with a median age of 62.5±8.2 years, most (96%) of whom were men with a median education (IQR) of 12 (12, 14) years. The median MELD-Na was 11 (8, 16) with most patients having a diagnosis of metabolic fatty liver disease (32%) followed by HCV (25%) and alcohol (18%). A third had prior OHE. 42% had ascites, 34% had varices and 24% were on non-selective beta-blockers (NSBB, Table S1).

CHE on QuickStroop was found in 126 (50%) patients. Patients with CHE on Quickstroop were more likely to be men who were older, with a lower educational attainment, worse cirrhosis severity and prior OHE (Table S1). NSBB use and presence of varices were similar between groups. Patients were followed for a median of 7 (IQR (3, 34)) months from testing. The median time-to-event for those who developed OHE was 6 (IQR (1, 24)) months, OHE-related hospitalizations was 3 (1, 20) months and all-cause hospitalizations was 7 (2, 18) months post- testing. Overall, 16% patients developed OHE, 10% needed OHE-related hospitalizations, 60% experienced all-cause hospitalizations, 22% were transplanted and 23% died during the observation period. More patients with CHE developed all these outcomes except all-cause hospitalizations on crude analysis (Table S1). Of those who died 37 (65%) were free of OHE and 44 (77%) were free of OHE-related hospitalizations before death. Of those who were transplanted 44 (81%) did not develop OHE and 47 (87%) were free of OHE-related hospitalizations before transplant. This pattern was also seen in time-to-event analysis for all outcomes. However, after adjustment for demographics, medications, cirrhosis severity and complications, CHE on Quickstroop remained significantly linked with lower time to OHE development and OHE related hospitalizations but not death or all-cause hospitalizations regardless of Cox proportional hazards or competing risks analyses(Figure 1 and Table S2). The variables that remained close to significant were MELD-Na, and prior OHE. Similar results were seen even when prior OHE patients were excluded (Cox HR,4.34, 95% CI 1.56–12.08, p=0.005 and competing risks HR3.79, 95% CI:1.36–10.60, p=0.011).

Figure:

Cox Proportional Hazard for Time to Overt Hepatic Encephalopathy Development. Time in months in the X-axis while Y-axis represents percent of patients free of overt hepatic encephalopathy (OHE). Covert hepatic encephalopathy (CHE) was diagnosed using QuickStroop. Red lines are those with CHE and blue are those without CHE.

In this prospective study of 250 outpatients with cirrhosis we found that CHE on QuickStroop increased the risk of outcomes specific to OHE (episodes and hospitalizations), which are clinically significant for patients and families. This finding remained consistent despite statistically controlling for demographics, cirrhosis severity, prior OHE, and medications. However, the initial association observed in the univariable analysis between CHE on Quickstroop and all-cause hospitalizations along with mortality, did not persist after adjusting for other clinical factors such as MELD-Na score and prior OHE. These findings indicate that even though CHE rates increased with advancing liver disease severity, the predictive potential is more specific for OHE-related complications. This specificity is important for risk-assessment and counseling patients and family members for OHE-related complications. Quickstroop primarily tests for reaction time and psychomotor speed, which may be adequate compared to the entire EncephalApp that also studies cognitive flexibility. These results extend the prior cross-sectional data.

While Quickstroop offers advantages such as being short and availability of US-based norms (at www.encephalapp.com), this assessment requires equipment and may not be suitable for patients with red-green color blindness, or those unfamiliar with mobile applications. Moreover, our population had mostly older men and included patients with prior OHE, these findings need external validation. We also did not compare Quickstroop to other modalities such as the psychometric hepatic encephalopathy score or the animal naming test. However, we demonstrated the potential of Quickstroop as a standalone test for predicting clinically relevant outcomes^{9, 10}.

Our data shows that Quickstroop demonstrates specificity for OHE-related outcomes in outpatients with cirrhosis. These findings can be used to counsel patients and families, improve precision of OHE prediction, and potentially encourage CHE therapy.