Association between vasoactive–inotropic score, morbidity and mortality after heart transplantation

Despite major advances in durable mechanical circulatory support (MCS), heart transplantation (HTx) remains the most valuable therapeutic option for patients with end-stage heart failure [1, 2].

inotropic support and the occurrence of septic shock, ventilator-associated pneumonia, bloodstream infection or renal replacement therapy.
RESULTS: A total of 151 patients underwent heart transplantation and admitted to intensive care unit. The median VIS max was 39.2 (interquartile range = 19.4-83.0). VIS max was independently associated with 1-year post-transplant mortality, as well as recipient age [hazard ratio (HR) = 1.004, P-value = 0.013], recipient gender (female to male: hazard ratio = 2.23, P-value = 0.047) and combined transplantation (hazard ratio = 2.85, P-value = 0.048). There was a significant association between VIS max and duration of mechanical ventilation (P-value < 0.001), length of stay (P-value = 0.002), duration of infused inotropes (P-value < 0.001), occurrence of bloodstream infections, septic shocks, ventilation-acquired pneumonia and renal replacement therapy.
CONCLUSIONS: VIS max calculated during the first 24 h after postoperative intensive care unit admission in transplanted patients is independently associated with 1-year mortality. In addition, length of stay, duration of mechanical ventilation and infused inotropes increased with increasing VIS max .

INTRODUCTION
Despite major advances in durable mechanical circulatory support (MCS), heart transplantation (HTx) remains the most valuable therapeutic option for patients with end-stage heart failure [1,2]. More than 6000 HTx procedures are performed each year worldwide, with a median post-transplant survival >12 years [3][4][5]. Primary graft dysfunction (PGD) is common complication after HTx and represents the leading cause of early post-transplant mortality [6,7]. International guidelines introduced the first consensual definition of the severity of PGD using a four-level scale: mild, moderate, severe PGD-left ventricle (PGD-LV) and PGDright ventricle [8]. In this grading scale, inotropic/vasopressors requirements were summarized as a dichotomous variable with a threshold defined arbitrary, thereby neglecting a part of the spectrum of inotropic requirements after HTx. Whether the detailed evaluation of inotropic requirements is independently associated with post-transplant outcomes on top of the current PGD classification remains unclear. The vasoactive-inotropic score (VIS) is a score summarizing the level of inotropic/vasopressors requirements calculated according to the type and doses of drugs [9]. It is usually calculated as a mean of inotropic support during a prespecified period. This score has been associated with prognosis in multiple situations, including cardiac surgery [10][11][12][13], but data are scarce after HTx [14,15]. Recently, VIS has been shown to be associated with poor short-term outcomes after paediatric HTx, but not independently associated with mortality [14]. The VIS has recently been updated as VIS max (defined by the maximum doses of inotropes/vasopressors received during the first 24 h). VIS max appeared to be a more valuable parameter after cardiac surgery [13,16], but has not yet been evaluated in adult patients undergoing HTx. We aimed to determine whether postoperative VIS max (i) was independently associated with post-transplant mortality and morbidity and (ii) could refine the prognostic value of the current international PGD classification.

Study design and population
We conducted an observational single-centre retrospective study in the Surgical ICU of Cardiology Institute at La Pitié -Salpê triè re University Hospital (Paris-France). Patients admitted to surgical ICU after HTx between January 2015 and December 2018 were included. Data were anonymously collected using computerized medical records.

Outcomes measures
The primary outcome was all-cause mortality at 1-year posttransplant. Secondary outcomes included number of days free from mechanical ventilation and inotropic support during the first 28 days post-HTx, and the occurrence of bloodstream infection, septic shock, ventilator-associated pneumonia (VAP) or renal replacement therapy (RRT).

Treatment protocols
Heart implantation was performed orthotopically using bicaval technique. The graft ischaemic time was defined as the time interval between the application of aortic cross-clamp in the donor and removal of aortic cross-clamp in the recipient. Celsior solution was used in donor heart preservation. Immunosuppression after HTx was administered based on a standard protocol described previously [17]. PGD is defined as severe ventricular dysfunction of the donor graft which fails to meet the circulatory requirements of the recipient in the immediate post-transplant period. It differs from secondary graft dysfunction, which is when a discernible cause for allograft dysfunction is identified. The monitoring and management of PGD is detailed in Supplementary Material, Methods.

Database and data collection
Baseline clinical data on donors and recipients from Paris were obtained from the national registry CRISTAL (Agence de la Biomé decine, French National Agency for Organ Procurement). Anonymized data from this registry are prospectively entered by dedicated staff at specific time points for each patient (day of listing, day of transplant) and are updated annually thereafter. This database is regularly audited. The extraction was carried out on 1 November 2021. We collected detailed characteristics of recipients and their donors including preoperative variables, postoperative variables in the ICU, postoperative variables during the first-year post-transplant (Supplementary Material, Methods).

Data availability statement
In accordance with the regulations, the data are stored locally in a secure database and are accessible in case of need to provide more details.

Statistical analysis
Categorical variables were described by frequencies and compared by chi-squared tests. Continuous variables were described by their means (± standard deviation) or medians [± interquartile range (IQR)] and compared Kruskal-Wallis non-parametric test. Cumulative survival curves for the time-to-event analyses were constructed according to the Kaplan-Meier method and compared with the log rank test.
Primary outcome. Cox regression was used to evaluate the association between VIS max groups (<16th, 17th-50th, 51th-83th and >84th percentile) and all-cause mortality during the first-year posttransplant. Candidate factors were selected when univariate likelihood ratio test P-value was <10%. Descending selection was then used and the final multivariable model selected factor significant at 5%. The additional prognosis value of VIS max on top of current PGD classification was assessed by analysing discrimination (comparison of concordance statistic) and reclassification indices [net reclassification improvement (NRI) and integrated discrimination improvement (IDI)]. The discrimination of the models was evaluated with the concordance statistic and were compared using the STATA's 'somersd' package. To evaluate the robustness of the association between VIS max and mortality, 200 bootstrap samples were generated (random sample with replacement, stratified by type of circulatory support at transplant); in each sample, a multivariable Cox model was generated using the same methodological approach described earlier.
Secondary outcomes. The association between VIS max and (i) binary outcomes (septic shock, VAP, bloodstream infection or RRT) was analysed using logistic regression and (ii) the number of days alive free from ventilation or inotropes during the first 30 days post-HTx by comparing and analysing the trend of number of days across VIS max categories using the non-parametric Kruskal-Wallis and the Cuzick test for trend, respectively.
A P-value <0.05 was considered statistically significant. Statistical analyses were performed with STATA/SE 14.2 (StataCorp, Texas, USA).

Study population
Between January 2015 and December 2018, 151 patients were transplanted and admitted to surgical ICU. Demographic and clinical characteristics of the population are described in Table 1. Vasoactive-inotropic score.

Primary end point-1-year mortality
In univariable analysis, among the 21 potential predictive variables analysed, 7 were associated with 1-year post-transplant mortality, including 3 recipients characteristics (age at transplant; gender; past history of cardiac surgery); 1 donor variable (age > _50 years old); and 4 transplant characteristics (combined transplantation; VIS max; category of VIS max ; and category of PGD) ( Table 2).

Risk stratification
We analysed the improvement in stratification of the risk of death during the first-year post-transplant when adding VIS max on top of the current ISHLT PGD classification. The discrimination of PGD classification alone, as assessed by the c-statistic, was 0.64 (95% CI = 0.55-0.73). The addition of VIS max significantly improved discrimination (c-statistic = 0.72, 95% CI = 0.62-0.82, P = 0.06). Reclassification indices were also significantly improved. The non-event NRI was 0.227 and the event NRI was 0.243. Global NRI was 0.47 representing a significant improvement in reclassification (P = 0.03). Similarly, the analysis of IDI revealed a significant improvement in reclassification (IDI = 0.043, P = 0.04).

Secondary end points: early morbidity
We found a significant association between VIS max and various post-transplant events during the first 30 days post-transplantation, including number of days alive free from infused inotropes (P-value < 0.001; Fig. 1A), number of days alive free from mechanical ventilation (P-value < 0.001; Fig. 1B), ICU LOS (P-value = 0.002), occurrence of bloodstream infections, septic shocks, VAP and RRT (P-value < 0.001 for all comparisons).

Internal validation
Among 200 bootstrap samples randomly generated for internal validation procedures, VIS was independently associated with 1- Weight mismatch was calculated as follows: (donor weight -recipient weight)/recipient weight Â 100. Weight mismatch was considered when the result was lower than -30%. BMI: body mass index; CI: confidence interval; HR: hazard ratio; MCS: mechanical circulatory support; PGD-LV: primary graft dysfunction-left ventricle; PGD-RV: primary graft dysfunction-right ventricle; Pre-HTx ECMO: pre-heart transplantation extracorporeal membrane oxygenation; VIS max : vasoactive-inotropic score (calculated with maximum doses 24 h postoperatively).
year all-cause mortality in 72% of the samples. VIS was the most selected variable, ahead of recipient age (68% of samples) and recipient sex (58% of samples). These results remained stable when forcing the variable 'PGD' into the automatically generated multivariable models.

DISCUSSION
In this study conducted in a contemporary cohort of 151 heart transplant recipients, VIS max calculated in the first 24 h after ICU admission was independently associated with morbidity and 1year mortality and significantly improved risk stratification of early death on top of the current ISHLT PGD classification.
When it comes to adult HTx, most studies concentrated on scores based on preoperative factors. Barge-Caballero et al. [15] calculated preoperative VIS and noted a continuous association between preoperative VIS and post-transplant mortality. Only 1 retrospective study by Venema et al. [18], involving 81-transplanted patients, calculated postoperative inotrope score and reported that it was significantly associated with 5-year survival.
In line with their results, VIS max independently predicted 1-year mortality in our work albeit some discrepancies between the studies. First, we found that PGD was common after HTx and frequently required ECMO. Whether inotrope score remained discriminant in a population under VA-ECMO was unclear so far. Venema et al. [18] reported a significant interaction between VIS max and MCS but the study was underpowered to assess this subgroup. Our data suggest VIS max still accurately predict outcomes for severe PGD requiring MCS. The high rate of VA-ECMO after HTx in our centre might be explained by local protocol and is supported by recently published data stating that VA-ECMO for HTx does not burden survival but could prevent from acute cardiovascular collapse and multiple organ failure [17]. Second, we used VIS max calculated during 24 h after HTx (versus mean inotrope score calculated during first 48 h for Venema et al.), which prevents from effective between-studies comparison. We considered VIS max as it could be easily translated into clinical practice and can provide early prognostication.
In a previous study involving >3000 patients, the predictive accuracy of VIS max was evaluated for a composite outcome, which included 30-day mortality, mediastinitis, stroke, acute kidney injury and  myocardial infarction in conventional cardiac surgery [13]. They reported VIS max predicted a composite of unfavorable outcomes and mortality up to 1-year after surgery. Median VIS max was 4.0 (IQR = 0.0-14.6), contrasting with our results. Heart transplant was associated with elevated VIS max mainly derived by PGD in our cohort. It is valuable to notice that survival curves (Fig. 2) split between 0 and 3 months but remained parallel afterward, suggesting VIS max rather correlates with mid-term outcome. Initial exposure to inotrope or vasopressors has no predictive value for long-term survival once patients discharged alive from hospital. In the literature, VIS was also an independent predictor of in-hospital mortality after left ventricular assist device when calculated after stabilization in the operating room and upon arrival to ICU [19]. In paediatric HTx, death tended to be higher in those with persistent high VIS until 48 h postoperatively, although not statistically significant [14]. Altogether, VIS max could be considered as a relevant stratification score for comparing transplanted patients.
In addition, this study showed that adding VIS max to the current ISHLT PGD classification significantly improved risk stratification after HTx as illustrated by the global net reclassification index of 0.47. However, discrimination of the model remained limited and not clinical relevant for individual risk stratification. In their study, Tadros et al. [20] noted that for VIS, the ROC curve exhibits 10 as the ideal cut-off with area under the curve >0.8 for PGD.
Our study also showed, in accordance with other studies, that higher VIS max was significantly associated to number of days free from mechanical ventilation and infused inotropes in ICU, and ICU LOS [14,16,19]. Moreover, risk for bloodstream infections, septic shock and VAP appears constant among VIS max percentiles. In fact, McIntosh et al. [21] showed that, in paediatric sepsis, VIS was independently associated with ICU LOS, ventilator days, and cardiac arrest/ECMO/mortality. The increased risk of infection along with increasing VIS max could be explained by the frequent use of mechanical, respiratory, and circulatory supports in severe patients, who were prevalent in our study, as well as by immune dysfunction secondary to circulatory failure and/or excess of inotropes. Pons et al. [22] have shown that in patients who had HTx, postoperative ECMO was an independent risk factor for early non-viral infection. Finally, VIS max increments were linearly associated with increased HR for RRT requirement after HTx as was also demonstrated by other studies [18,23]. Hou et al. [24] showed that VIS max was an independent predictor of postoperative acute kidney injury in adult patients after cardiovascular surgery and increased prognostic accuracy of Society of Thoracic Surgeons score, allowing a risk reclassification. Hence using VIS max as an indicator for renal aggression could help implement early nephroprotective measures.

Limitations
This study has 4 main limitations. First, the small sample size with limited number of 'events' such as death reduced the power of statistical analysis. It was reflected in the wide confidence interval of OR in the cox model. Second, this is a single-centre retrospective study. Despite being retrospective, there were few missing data, but it lacks an external validation cohort. Third, 1 maximum value of inotropic and vasoactive support may be subjected to confounding factors such as momentary deeper sedation, postoperative bleeding/hypovolaemia, or procedures requiring temporarily increased pharmacological cardiovascular support. Albeit VIS max appears easier for clinical practice, whether total cumulative inotrope and vasoactive exposure during the first 24 h in ICU could better predict outcomes remains unknown. Finally, the population mainly composed of severe PGD, might burdened external validity of the study. The high prevalence of severe PGD in French cohorts compared to literature might be explained by donors' higher mean age (48 ± 14.4 years old), previously published as a major risk factor for PGD [25]. It is valuable to highlight those results might have been influenced by local practice and centre experience as our hospital is referred for VA-ECMO and trained medical, surgical and paramedical teams significantly reduce VA-ECMO related complications [26,27]. However, exposure to VA-ECMO did not change the interpretation of the results and VIS max remained reliable even among severe PGD patients.

CONCLUSION
To our knowledge, this is the first study describing adverse outcomes and mortality after adult HTx based on post-transplant VIS max . It confirms that VIS max calculated during the first 24 h after postoperative ICU admission is independently associated with 1-year mortality. In addition, ICU LOS, duration of mechanical ventilation and infused inotropes increased with increasing VIS max . Future prospective multi-institutional studies are necessary to better evaluate the utility of postoperative VIS max after HTx.

SUPPLEMENTARY MATERIAL
Supplementary material is available at ICVTS online.

Funding
None.