Citrulline Trajectory during Allogeneic Hematopoietic Stem Cell Transplantation Is Correlated to Conditioning Intensity and Non-Relapse Mortality

Introduction

Citrulline, a non-essential amino acid produced exclusively by enterocytes in the small intestine and involved in the synthesis of L-arginine, is not metabolized by the liver. Therefore citrulline serum concentration is highly correlated with functional enterocyte mass, and decreases with digestive toxicity induced by conditioning therapy (radiotherapy and/or chemotherapy) for hematopoietic stem cell transplantation (HSCT) .

Acute Graft-versus-host disease (GvHD), one of the major complications of HSCT, is correlated to conditioning-induced gut barrier damage and may be predicted by pre-transplant serum citrulline level (Rashidi, BBMT 2018). It could be interesting to know whether citrulline kinetics could also represent a biomarker for conditioning toxicity, non-relapse mortality (NRM), and GvHD. The aim of this study is thus to define group-based trajectory modeling, to identify clusters of individual serum citrulline kinetics in the early phase of allogeneic HCST, and to test whether these unsupervised trajectories were correlated with these early complications.

Materials and Methods

Serum citrulline was quantified by liquid chromatography in blood samples collected from consecutive patients who received an allogeneic HSCT in our institution between July 2014 and November 2019. These samples were drawn at different time-points: pre-transplant (D-7, D-3); day of transplant (D0), and post-transplant (D7, D15, D21).

Distinct trajectories were identified for serum citrulline by using the semiparametric mixture model described by Nagin (Nagin, Stat Methods Med Res 2018).

Results

Among 161 patients (pts) included in the study, with a median age of 53 years (17-72), 98 pts (60.9%) received a reduced-intensity conditioning (RIC), 36 pts (22.4%) reduced-toxicity conditioning (RTC), 18 pts (11.1 %) sequential conditioning, and 9 pts (5.6%) myeloablative conditioning (MAC). Donor were identical sibling (22%), matched unrelated donor (52%) and haploidentical sibling (25%). Graft source was peripheral blood mononuclear cells in 144 pts (89.4%) and bone marrow in 17 pts (10.6%) respectively. HCT-CI score was low, intermediate and high-risk in 38%, 32%, and 30% of pts respectively. Disease-Risk Index (DRI) was low/intermediate in 111 pts (69%) and high/very-high in 50 pts (31%).

With a median follow up of 29.1 month, 3-year overall survival (OS), disease-free survival (DFS), and NRM rates were 64.5%, 58.3%, and 18.9%, respectively.

The median number of citrulline samples per patient was 7 [3-16]. Median citrulline concentrations before conditioning and at D-3, D0, D7 and D15 were statistically different during RIC, RTC, MAC, and sequential conditioning (p<0,001 respectively) but was not different at D21 (p=0.296).

In the whole cohort, 3 citrulline trajectories were determined in an unsupervised method. Patients belonging to these 3 trajectories were different according to intensity of conditioning received with lower citrulline trajectories during MAC and sequential conditioning (p<0.001). In the uper citrulline trajectorie, pts were significantly older (p=0.005). However, citrulline trajectories were not correlated to OS (p = 0.1), NRM (p=0.24), cumulative incidence of acute GvHD (p=0.39) or chronic GvHD (p=0.2).

After restricting the analysis to pts who received RIC conditioning (n=98), higher pre-HSCT citrulline concentrations were associated with a lower NRM (p=0.042). Unsupervised analysis in this setting individualized 4 clusters of individual trajectories (figure 1), that did neither distinguish age (p=0.28), DRI (p=0.87), HCT-CI score (p=0.81) nor the incidence of acute (p=0.6) or chronic (p=0.4) GvHD. However, the lowest citrulline trajectory contained significantly more haploidentical transplantations (p=0.004) and less pts who received antithymocyte globulin for GvHD prophylaxis (p=0.005). Interestingly in this RIC cohort, cumulative incidence of NRM at 12 months was 23%, 21%, 8%, and 0% respectively according to the 4 citrulline trajectories (figure 2).

Conclusion

In patients receiving allogeneic HSCT, the variation of serum citrulline concentrations depends on the intensity of the conditioning regimen. In patients who received RIC conditioning, lower plasma citrulline trajectories are associated with higher NRM. In this setting, citrulline may be an attractive biomarker for predicting conditioning toxicity and NRM.