Pretransplant Lymphopenia is a Novel Prognostic Factor in CMV and Non-CMV Invasive Infection After Liver Transplantation

Abstract

Infection following liver transplantation (LT) remains a leading cause of morbidity and mortality. Risk of infection after LT is highest in those who are most immunosuppressed; yet, to date no standard blood marker of one’s degree of immunosuppression or risk index has been established. The purpose of this study was to determine if pretransplant lymphopenia (absolute lymphocyte count < 500 cells/mm3 within 24 hours before LT) is a candidate marker of immunosuppression and useful in predicting risk of cytomegalovirus (CMV) disease and non-CMV invasive infection after LT. Data was extracted from medical records for all primary, solitary liver transplants performed at Tufts Medical Center from 1999–2009. 276 patients had sufficient data to be included in the analysis. Of these, 52% developed CMV or non-CMV invasive infection within 5 years of LT. By 2 years, 23 (8%) had CMV disease and 103 (37%) at least one non-CMV invasive infection. More lymphopenic than non-lymphopenic patients developed CMV (21% versus 4%, P < 0.0001) and non-CMV invasive infection (50% versus 33%, P = 0.02). In multivariable survival analysis, pretransplant lymphopenia was the strongest independent predictor of CMV disease (hazards ratio [HR] 5.52, 95% confidence interval [CI] = 2.31–13.1; P = 0.001) after adjustments for known risk factors, including CMV serostatus (HR 4.72, 95% CI = 2.01–11.1; P < 0.0001). Both pretransplant lymphopenia (HR 1.64, 95% CI = 1.14–2.53; P = 0.03) and CMV (HR 2.93, 95% CI = 1.23–6.92; P = 0.02) independently predicted non-CMV infection. Our results suggest that pretransplant lymphopenia is a novel independent predictor of both CMV disease and non-CMV invasive infection after LT and a candidate marker of immunosuppression in LT recipients.

Background

Infection is a primary source of morbidity and mortality following liver transplantation (LT). It is the cause of death in one third of solid organ transplant (SOT) recipients and experienced by the majority of LT recipients.¹ Risk of infection after SOT is highest in those who are most immunosuppressed, yet, to date no standard blood marker of one’s degree of immunosuppression or risk index has been established.² Routine evaluation of total white blood cell and neutrophil counts, both widely studied as predictors of infection in immunocompetent and immunocompromised hosts, was adopted in the 1960’s.³ The lymphocyte count as a candidate marker of immunosuppression, however, has not commanded such attention.

From HIV immunologic studies, we know that CD4 counts are reliable markers of net immunosuppression and it is standard of care to use them clinically to guide initiation of prophylactic antimicrobial agents. Similar strategies have not been adopted in posttransplant monitoring although a few transplant centers have studied the utility of the T cell and its subsets as well as other potential markers of net immunosuppression including the Immunknow™ test (Clinimmune Labs, Aurora CO) and immunoglobulin levels.⁴ None of these markers, however, have proven useful in predicting infection after SOT. More recent studies have therefore explored the clinical utility of cytomegalovirus (CMV)-specific cell-mediated immunity after SOT, but have yet to relate such immunologic monitoring to other infectious events following transplantation.^{5, 6} Moreover, immunoglobulin levels and specific cell mediated markers have long processing times and are costly, limiting their utility.

A more rapidly available, less expensive tool is needed to impact real-time clinical decision making in our SOT hosts similar to the utility of the absolute neutrophil count in bone marrow transplant patients and CD4 count in HIV patients. We believe the total lymphocyte count has potential to similarly guide prophylactic and preemptive antimicrobial therapy and surveillance strategies in SOT recipients. Several studies have suggested that low lymphocyte counts indicate inadequate reserve to mount an appropriate immunologic response^.7 Clinically, the total lymphocyte count is useful in monitoring response to antiretroviral therapy⁸ as well as antitumor immune responses.⁹ Impaired function of CD8⁺ and CD4⁺ T cells and peritransplant lymphopenia have been correlated with decreased survival in hepatocellular carcinoma and several other malignancies.^{10, 11}

Still, few have considered the relevance of lymphocyte counts for identifying and monitoring patients at risk for posttransplant related infection. One group in Spain studied pretransplant lymphocyte counts and suggested reduced T cell subset and total lymphocyte counts preceding LT predicted risk of developing any type of infection following LT.¹² Our primary objective was to further explore the total lymphocyte count as a potential marker of immunosuppression in LT recipients in hopes of elucidating a reliable and more readily available marker of immunosuppression for all SOT recipients.

PATIENTS AND METHODS

Study Design and Patient Selection

The study group consisted of consecutive patients undergoing primary LT at our institution from January 1999 to December 2009. Adult and pediatric patients with complete data sets who survived at least 24 hours after LT were included in the analysis. Dialysis-dependent and combined liver-kidney transplant recipients were excluded to reduce confounding effects of decreased lymphocyte counts in end stage renal disease.¹³

Patients were followed until December 2011 (at least 2 years per patient), death, or retransplantation. For patients with acute primary graft failure who received a second LT within 30 days of primary LT, the two transplants were consolidated into one event with pretransplant data collected before the first LT. If one donor was CMV seropositive (D+) and the other CMV seronegative (D−), the donor (D) CMV serostatus was considered positive (D+). Any other second transplant was considered a second event and data censored at time of second LT.

Data was retrospectively collected from patients’ electronic and paper medical records. To limit bias, baseline data was collected by one research team member and outcome data by a different team member, in a separate database, blinded to baseline lymphocyte counts and other clinical data. The datasets were de-identified and merged together for analysis.

Approval was obtained from the Tufts Medical Center Institutional Review Board.

Immunosuppression

Standard initial immunosuppressive regimen consisted of cyclosporine or tacrolimus, plus azathioprine or mycophenolate mofetil (MMF), and corticosteroids. Patients with end stage renal disease (CKD-EPI eGFR ≤ 30 ml/min)¹⁴ received antithymocyte antibodies (ATG) - muromonab-CD3, daclizumab, basiliximab, or thymoglobulin - during induction, in lieu of tacrolimus. A calcineurin inhibitor was then added shortly after transplantation. Rejection episodes were uniformly treated with intravenous methylprednisone for five days and, if no response achieved, with ATG.

CMV Prophylaxis Regimens

From 1999 until 2003, patients received 90 days of oral ganciclovir (one gram three times daily or the appropriate dose adjusted for impaired renal function).¹⁵ Additionally, high-risk patients (D+/recipient negative [R−]) received seven doses of intravenous CMV immunoglobulin (CMVIgG). After 2004, all patients received 90 to 120 days (longer in high risk patients) of oral valganciclovir (900 mg once daily or the appropriate dose adjusted for impaired renal function). CMVIgG was not routinely given in this later era. Duration of prophylaxis was recorded as: total number of days after LT to cessation of prophylaxis minus number of days prophylaxis was interrupted for any reason (usually for leukopenia, neutropenia or decreased renal function).

Definitions

Pretransplant lymphopenia (or neutropenia) was defined as an absolute lymphocyte (or neutrophil) count (ALC or ANC) of less than or equal to 500 lymphocytes or neutrophils/mm³ within the 24 hours before LT surgery. Pretransplant leukopenia was defined as a total white blood cell count (WBC) of less than or equal to 3000 leukocytes/mm³ in the 24 hours prior to LT.³

Severe Liver Disease was defined by the designation of Status I, the grading system used until 2002, or subsequently, as having a Model for End Stage Liver Disease (MELD) score of at least 24 at the time of transplantation.¹⁶

Definite liver allograft rejection was defined by histological evidence of endotheliitis with portal tract expansion by mononuclear cells and infiltration and swelling of bile ducts. Probable rejection was defined as resolution of hyperbilirubinemia and transaminitis following pulsed steroid treatment, with or without ATG, in the absence of liver biopsy confirmation when no other cause of liver dysfunction was identified.¹⁷

CMV viral replication was tested in blood buffy coat, bone marrow aspirates, and tissue biopsies via the rapid shell-vial technique¹⁸ and conventional viral culture. Molecular methods of detection were used on blood buffy coats: the Hybrid Capture CMV DNA Assay (version 2.0, Digene Corporation, Silver Spring, Maryland)¹⁹ until 2008 and a PCR-based assay from 2008–2009 (Quest Diagnostics, Chantilly, Virginia).²⁰ Biopsy material was examined histologically for characteristic CMV-induced changes and immunologically stained for CMV inclusions.

CMV disease was defined as organ damage or systemic illness and detection of cytomegalovirus via the above mentioned histological, microbiological, or molecular methods.²¹ CMV infection (viral replication without organ damage or symptomatology) was not a CMV disease event in these analyses.

Non-CMV invasive infection was defined as the presence of a clinical syndrome or end organ damage in conjunction with isolation of a pathogenic micro-organism compatible with disease at that site including: bacteremia as well as invasive fungal, mycobacterial, and non-CMV viral infections. Recurrent hepatitis B or C was not counted as an infectious event in this study.

Standard definition by Munoz-Price et al¹⁷ was used to define blood stream infections (BSI). Characterization of blood cultures containing bacteria that typically colonize the skin as a true BSI, rather than contamination, required: two consecutive positive blood cultures, two positive blood cultures within 72 hours, or one positive blood culture and one positive intravascular catheter tip culture within 72 hours.²²

Invasive fungal infection (IFI) was defined as the identification of fungal or yeast species by culture or histological exam from a normally sterile site in the setting of a clinical syndrome or end organ damage.²³ Solitary positive sputum, biliary tube, urine, or Foley catheter cultures were not IFI events in these analyses.

Outcomes

The clinical outcomes of LT patients with and without pretransplant lymphopenia were determined within 2 years of LT. Ancillary analyses employed a 5-year follow-up period. Primary outcome was time to CMV disease. Secondary outcomes were time to first non-CMV invasive infection and to death.

Predictor Variables

Potential predictors of primary and secondary outcomes were organized into pretransplant (preLT), intraoperative, and posttransplant (postLT) factors. PreLT factors were demographic: transplant age, gender, and race; and donor and recipient characteristics: severity and etiology of liver disease; comorbid conditions; list time; early (1999–2003) versus late (2004–2009) transplant era; baseline: renal (creatinine and calculated eGFR), hematologic (WBC, ANC, ALC), and nutritional status (albumin); CMV D/R serostatus; ABO D/R compatibility; donor type (live versus deceased); cold ischemia time; and data for donor risk index (DRI) which incorporated donor age, cause of death, race, height, split versus whole liver, and cardiac death and was calculated according to Feng.²⁴ Intraoperative factors included: volume (in units) of blood products transfused and type of biliary anastomosis. Postoperative categorical variables were: induction and maintenance immunosuppressive regimens, ATG as part of induction, and type of CMV prophylaxis. Additionally, postLT events occurring at varying times after LT were evaluated as time-dependent covariates and included time (in days after LT) to: completion of CMV prophylaxis, rejection, treatment of rejection, infection and last live follow up.

Statistical analysis

Patients were divided into one of two groups: those with pretransplant lymphopenia and those without. Data are shown as the mean (or median) ± standard deviation (or interquartile range [IQR], Q_{1 –} Q₃). Chi-square and Fisher’s exact tests were used to assess relationships between categorical variables and Student’s t and Wilcoxon-Mann-Whitney tests for comparisons of continuous variables.²⁵ Pretransplant lymphopenia as a predictor of primary and secondary outcomes was preliminarily assessed using the Kaplan-Meier method and compared with the log-rank test.²⁶

Survival analysis was used to identify risk factors associated with CMV and non-CMV invasive infection and death.^{27, 28} All potential predictors identified in univariate analyses (P < 0.10), as well as other clinically relevant factors considered potential confounders by related literature, were entered into multivariable models using backward selection to assess their effect.²⁹

Testing of proportional hazards was performed. When the assumption of proportional hazards was violated and a variable that occurred at the time of transplantation had a hazards ratio that changed over time, this variable was modeled as a time-dependent covariate to account for the variation in risk.²⁸ Other time-dependent covariates were the postLT events listed above as well as (1) CMV disease in the non-CMV invasive infection models, (2) non-CMV in the CMV models and (3) both CMV and non-CMV events in mortality models. Time-dependent events that occurred after the outcome of interest did not enter the analyses.

Multivariable analyses of primary and secondary outcomes were performed with competing risks models. The competing risk was all-cause mortality. P values ≤ 0.05 (two-sided) were considered statistically significant. Hazards ratios (HR) are reported with 95% confidence intervals (CI). Analyses were performed using SAS, Version 9.3, (SAS Institute, Cary, North Carolina).

RESULTS

Patients

During the study period, 313 patients underwent primary LT. Thirty seven patients were excluded according to criteria for a final cohort of 276 patients for analysis, Figure 1. Of these 276 patients, 16 were retransplanted during the study period: 9 for acute primary graft failure (consolidated into 1 event; censored at time of death or minimum 2-year follow up) and 7 for other causes more than 30 days apart (transplants not consolidated; censored at time of second LT). Median and mean follow up times for the cohort (n=276) were 5.1 years and 5.3 years ± 65 days, respectively.

Figure 1.

Construction of the Liver Transplant (LT) cohort.

Outcomes

Table 1 reports patient characteristics among the 276 patients, grouped by those with (n=68, 25%) and without (n=208, 75%) pretransplant lymphopenia. Proportionally more lymphopenic patients were also leukopenic, yet no patient in either group was neutropenic before LT. At LT, proportionally more lymphopenic patients were also older, P = 0.04, and had higher serum creatinine levels, P = 0.04, compared with non-lymphopenic patients.

Table 1.

Distribution of baseline study characteristics by pretransplant lymphopenia at time of liver transplant (LT) in n=276 LT recipients

		Pretransplant Lymphopenia
Characteristic	No. Patients (N = 276)	ALC≤500 (N = 68)	ALC>500 (N = 208)	P
Age at transplantation, median years (IQR)	51 (46 – 57)	54 (48 – 60)	51 (45 – 57)	0.04
Male gender	189 (69)	45 (66)	144 (69)	0.65
White race (vs. non-white)	239 (87)	62 (91)	177 (85)	0.23
Status I or MELD > 24	112 (41)	21 (31)	91 (44)	0.07
Early transplant era 1999–2003 (vs. 2004–2009)	180 (65)	45 (66)	135 (65)	0.88
Etiology of liver disease
Hepatitis B	23 (8)	5 (7)	18 (9)	0.81
Hepatitis C	114 ((41)	26 (38)	88 (42)	0.48
Hepatocellular Carcinoma	75 (27)	15 (22)	60 (29)	0.35
Alcoholic Cirrhosis	83 (30)	20 (29)	63 (30)	0.82
Autoimmune Hepatitis	12 (4)	5 (7)	7 (3)	0.17
Nonalcoholic Steatosis	6 (2)	5 (7)	1 (0)	1.00
Primary Biliary Cirrhosis	22 (8)	15 (22)	7 (3)	0.44
Primary Sclerosing Cholangitis	34 (12)	9 (13)	25 (12)	0.83
Idiopathic	24 (9)	9 (13)	15 (7)	0.14
Pretransplant conditions
Coronary Artery Disease	48 (18)	11 (16)	37 (18)	0.86
Diabetes, Type I or II	67 (24)	16 (24)	51 (25)	1.00
Creatinine, median mg/dL (IQR)	0.9 (0.8 – 1.3)	1.1 (0.8 – 1.5)	0.9 (0.7 – 1.2)	0.04
CKD-EPI eGFR, median mL/min (IQR)	94 (69 – 128)	83 (61 – 127)	96 (71 – 131)	0.06
CKD-EPI eGFR ≤ 30 mL/min*	17 (6)	6 (9)	11 (6)	0.38
Albumin, median g/dl (IQR)	2.7 (2.2 – 3.0)	2.6 (2.2 – 3.1)	2.7 (2.2 – 3.0)	0.73
WBC ≤ 3000 cells/mm3	45 (16)	21 (31)	24 (12)	0.002
ANC ≤ 1000 cells/mm3	0 (0)	0 (0)	0 (0)	--
Donor Characteristics
Donor Risk Index, median score (IQR)	1.26 (0.93 – 1.59)	1.19 (0.95 – 1.61)	1.22 (0.91 – 1.58)	0.35
Living (vs. deceased) donor	39 (14)	10 (15)	29 (14)	0.84
Cold ischemia time, median minutes (IQR)	382 (285 – 469)	389 (285 – 474)	420 (291–509)	0.24
ABO compatible (vs. incompatible)	249 (96)	63 (91)	186 (90)	0.46
CMV D/R serostatus
Recipient CMV +	150 (54)	31 (46)	119 (57)	0.12
Donor CMV +	126 (46)	37 (54)	89 (43)	0.12
D+/R−	57 (21)	20 (29)	37 (18)	0.06
D+/R+	69 (25)	17 (25)	52 (25)	1.00
D−/R−	69 (25)	17 (25)	52 (25)	1.00
D−/R+	81 (29)	14 (21)	67 (32)	0.09
Intraoperative Factors
Packed Red Blood Cells, median units (IQR)	7 (4 – 11)	8 (5 – 12)	7 (4 – 10)	0.07
Platelets, median units (IQR)	40 (6 – 68)	40 (8 – 80)	39 (6 – 63)	0.19
Fresh Frozen Plasma, median units (IQR)	20 (13 – 28)	20 (14 – 30)	20 (12 – 28)	0.37
Roux-en-Y anastomosis (vs.end-to-end)	83 (30)	21 (31)	62 (30)	0.88
Primary CMV prophylaxis regimen
Valganciclovir (vs. ganciclovir)	151 (55)	32 (47)	119 (57)	0.16
Primary CMV prophylaxis + CMVIgG (≥1 dose)	71 (26)	21 (31)	50 (24)	0.26
MMF-based immunosuppression (vs. non-MMF)	215 (78)	52 (77)	163 (78)	0.74
ATG induction (vs. none)	21 (8)	5 (7)	16 (8)	1.00

NOTE. Variables with P > 0.10 in univariate analysis (UV) and therefore not depicted: age, gender, race, etiology of liver diseases (except HCV as shown); comorbidities; list time; transplant era; baseline: WBC, ANC, creatinine, CKD-EPI eGFR, and albumin; intraoperative cryoprecipitate; ABO-compatible; CMV donor and recipient serogroups; cold ischemia time; type of donor; donor risk index; induction and maintenance immunosuppresion; and ATG induction.

^*Effect of Roux-en-Y anastomosis vs no Roux-en-y anastomosis in predicting non-CMV invasive infection varied over time and was modeled as a time-depentdent covariate to assess its effect (1) the first year of LT and (2) more than 1 year after LT. Roux-en-Y was predictive of non-CMV events within the first year of LT, but not predictive of non-CMV events that occurred more than 1 year of LT.

^§CMV disease, as a time-dependent covariate, was predictive of subsequent non-CMV events. Nonsignificant time-dependent covariates were duration of CMV prohpylaxis, rejection and treatment of rejection with ATG or steroids.

Abbreviations. ANC, absolute neutrophil count; ATG, antithymocyte globulin; CI, confidence interval; CKD-EPI eGFR, Chronic Kidney Disease Epidemiology Collaboration estimated Glomerular Filtration Rate; CMV, cytomegalovirus; HCV, hepatitis C disease; HR, hazards ratio; LT, liver transplantation; MELD, model for end stage liver disease; RBC, red blood cells; UV, univariate; WBC, total white blood cell count.

Primary and secondary outcomes in the first two years of LT, stratified by pretransplant lymphopenia, demonstrate a significantly larger proportion of lymphopenic than non-lymphopenic patients developed both CMV disease, P = 0.001, and non-CMV invasive infection, P = 0.02, while the groups did not differ with respect to the proportion of patients who died, P = 0.59, Table 2.

Table 2.

Outcomes after liver transplantation (LT) stratified by pretransplant lymphopenia in 276 LT recipients.

Outcome		Number with outcome (%)
	Total N = 276	ALC ≤ 500 N = 68	ALC > 500 N = 208	P-value
CMV Disease (at 2 years)	23 (8)	14 (21)	9 (4)	0.0001
CMV Disease (at 5 years)	26 (9)	14 (21)	12 (6)	0.001
Non-CMV Invasive Infection (at 2 years)	103 (37)	34 (50)	69 (33)	0.02
Bloodstream	71 (26)	24 (35)	47 (23)	0.05
Fungal	40 (15)	14 (21)	26 (13)	0.11
Viral*	17 (6)	6 (9)	11 (5)	0.06
Mycobacterial	2 (1)	2 (3)	0 (0)	--
Non-CMV Invasive Infection (at 5 years)	118 (43)	39 (57)	79 (38)	0.01
Bloodstream	82 (30)	27 (39)	55 (27)	0.07
Fungal	49 (18)	16 (23)	33 (16)	0.20
Viral*	25 (9)	11 (16)	14 (7)	0.03
Mycobacterial	2 (1)	2 (3)	0 (0)	--
Rejection (at 2 years)	117 (42)	27 (39)	90 (44)	0.58
Rejection (at 5 years)	130 (47)	33 (48)	97 (47)	0.89
Treated with ATG	10 (4)	1 (2)	9 (4)	0.46
Treated with steroids	110 (40)	27 (40)	83 (40)	1.00
CMV prophylaxis duration: median days (IQR)	91 (87 – 96)	91 (84 – 96)	91 (87 – 96)	0.81
Retransplantation (overall)	9 (3)	0 (0)	9 (4)	--
Death (at 2 years)	49 (18)	14 (20)	35 (17)	0.59
Death (at 5 years)	72 (26)	20 (29)	52 (25)	0.53
Follow-up time: median days (IQR)	1832 (856 – 3050)	1748 (647–3313)	1882 (884 – 2996)	0.93

^*Invasive viral infections other than CMV included: Epstein Barr viremia (EBV) or EBV-related posttransplant lymphoproliferative disorder, disseminated herpes simplex or varicella-zoster, as well as adenovirus (n=1), influenza type A or B (n=2), norovirus (n=1), parainfluenza (n=1), and parvovirus (n=1). Herpes virus limited to oral or genital mucosa was not counted as an invasive viral infection in this analysis.

Abbreviations. ALC, absolute lymphocyte count; ATG, antithymocyte globulin; BSI, blood stream infection; CMV, cytomegalovirus; EBV, (Epstein Barr virus); IFI, invasive fungal infection; IQR, interquartile range; LT, liver transplantation.

Lymphopenia and CMV disease

Within 2 years of LT, 23 (8%) of 276 patients developed CMV disease in a median of 142 days (IQR 127–177). Of 23 patients with CMV disease, 15 (65%) had received valganciclovir and 8 (35%) ganciclovir. All patients received primary CMV prophylaxis for a median of 90 days. All 23 CMV events occurred after cessation of prophylaxis in a median of 50 days post-discontinuation (IQR 40–155).

There was a significant difference in those with pretransplant lymphopenia (21%) compared to those without (4%) who developed CMV disease, P = 0.001, Table 2. While the time to develop CMV disease after LT did not differ significantly between the groups, there was a trend towards earlier occurrence of CMV in those with pretransplant lymphopenia (median 140 days) compared to those without (median 211 days), P = 0.06.

The Kaplan-Meier survival curves for time free from CMV disease stratified by those with and without pretransplant lymphopenia, diverged in a statistically significant way immediately following standard cessation of CMV prophylaxis, Figure 2A. Patients with pretransplant lymphopenia were more likely to develop CMV disease compared to patients without (log-rank test, P < 0.0001). Even after adjusting for CMV serostatus by restricting the cohort to at-risk patients (CMV seropositive donor and/or recipient), n=207, the risk of CMV disease remained statistically significant in those with pretransplant lymphopenia (log-rank test, P < 0.0001), Figure 2B.

Figure 2.

Kaplan-Meier estimates of survival free of CMV disease in the first 2 years after liver transplantation. Figure 2A, The entire cohort, n=276, by those with (dashed line) and without (solid line) pretransplant lymphopenia (ALC > 500 cells/mm3). The associated increased risk of CMV disease in patients with pretransplant lymphopenia is significant in the entire cohort, log-rank test P<0.0001. Figure 2B, The cohort restricted to at-risk CMV patients (CMV seropositive donor and/or CMV seropositive recipient), n=207, by those with (dashed line) and without (solid line) pretransplant lymphopenia. The probability of remaining CMV disease-free was significantly reduced in lymphopenic patients even when the cohort is restricted to at-risk CMV patients, log-rank test P<0.0001. Abbreviations. ALC, absolute lymphocyte count; CMV, cytomegalovirus.

Multivariable survival analysis confirmed the strong associated risk of CMV disease in patients with pretransplant lymphopenia (HR 5.52, CI 2.31–13.1; P = 0.001), independent of highest-risk CMV serostatus (HR 4.72, CI 2.01–11.1; P < 0.0001), Table 3. The only time-dependent variable associated with increased risk was other invasive viral infection; more intraoperative units of pRBCs was associated with reduced risk. Other potential risk factors explored in the models, including hepatitis C, DRI, induction and maintenance immunosuppression, rejection and its treatment with ATG or steroids, did not affect risk of CMV within 2 years of LT. Results were similar for CMV disease within 5 years of LT as the outcome (data not shown).

Table 3.

Univariate and multivariable risks for CMV disease in the first 2 years after LT in the total cohort, n=276.

Variable	Univariate		Multivariable*
	HR (95% CI)	P value	HR (95% CI)	P value
Pretransplant lymphopenia	5.32 (2.32–12.0)	<0.0001	5.52 (2.31–13.1)	0.001
Early era LT (1999–2003)	0.52 (0.22–1.13)	0.07	--	--
Coronary Artery Disease (CAD)	2.21 (0.94–5.32)	0.08	--	--
High risk CMV serostatus (D+/R−)*	4.52 (2.01–10.3)	0.0003	4.72 (2.01–11.1)	<0.0001*
Packed RBC (per intraoperative unit)	0.80 (0.64–0.93)	0.03	0.81 (0.61–0.92)	0.01
Other invasive viral infection§	4.02 (1.64–10.1)	0.004	3.33 (1.33–8.44)	0.02

NOTE. Variables with P > 0.10 in univariate (UV) analysis and therefore not depicted: age, gender, race, severity and etiology of liver disease, including hepatitis C; comorbidities (except CAD as shown); list time; pretransplant creatinine, CKD-EPI eGFR, CKD-EPI eGFR ≤ 30, WBC, ANC, albumin; ABO-compatibile; donor type; donor risk index; cold ischemia time; intraoperative platelets and cryoprecipitate; induction and maintenance immunosuppressive regimen; ATG induction.

^*CMV serostatus combinations not significant in UV analysis: D−/R−, D+/R+, D−/R+, or CMV R status; D+/R− (as shown) and CMV D+ (not shown) were significant in UV and multivariable (MV) analysis. These were run as separate survival models; D+/R− was used in MV model shown here. When CMV D+ was substituted in a separate MV model, HR was 6.73 (CI 2.01–22.9) and P = 0.002.

^§Other invasive viral infection was the only time-dependent covariate significant in UV and MV analysis; other time-dependent covariates tested and not significant included: composite non-invasive infection, BSI alone, IFI alone, rejection, treatment of rejection with ATG or steroids, and duration of CMV prophylaxis.

Abbreviations. ANC, absolute neutrophil count; CI, confidence interval; CKD-EPI eGFR, Chronic Kidney Disease Epidemiology Collaboration estimated Glomerular Filtration Rate; CMV, cytomegalovirus; D+, seropositive donor; D−, seronegative donor; R+, seropositive; R−, seronegative recipient; D, donor; HR, hazards ratio; LT, liver transplantation; MV, multivariable; R, recipient; RBC, red blood cells; UV, univariate; WBC, total white blood cell.

The relationship between the recovery, persistence and/or new appearance of lymphopenia at 30 days after LT and CMV disease was examined. Only patients with preLT lymphopenia had an elevated risk of CMV disease, regardless of the presence or absence of lymphopenia at 30 days after transplantation (data not shown).

Lymphopenia and non-CMV invasive infection

In the first 2 years after LT, 103 (37%) patients developed at least one non-CMV invasive infection. Table 2 depicts the distribution of infection type in those with and without pretransplant lymphopenia, including several patients who developed more than one type of invasive infection. The median time to first non-CMV infection was 23 days in lymphopenic compared with 39 days in non-lymphopenic patients, P = 0.18.

Pretransplant lymphopenia (HR 1.82, CI 1.22–2.73; P = 0.009) and CMV disease (HR 3.01, CI 1.32–7.31; P = 0.01) were two significant univariate predictors of non-CMV invasive infection after LT, Table 4. Severe liver disease and hepatitis C trended towards reduced risk of non-CMV invasive infection while ganciclovir and more units of intraoperative pRBCs and platelets increased risk. Additionally, because Roux-en-Y anastomosis violated the proportional hazards assumption it was treated as a time-dependent covariate and separate HRs were generated to assess its associated risk (1) within the first year and (2) more than 1 year after LT. Roux-en-Y anastomosis was associated with an increased risk of non-CMV infection during the first year only. After 1 year, the risk was the same in those who had and those who did not have a Roux-en-Y anastomosis (see also Table 4 footnote). All other variables were not significant in univariate analysis, including DRI, induction and maintenance immunosuppression regimens, and treatment of rejection with ATG or steroids.

Table 4.

Univariate and multivariable risks for non-CMV invasive infection in the first 2 years after LT in the total cohort, n=276.

Variable
	Univariate		Multivariable
	HR (95% CI)	P value	HR (95% CI)	P value
Pretransplant lymphopenia	1.82 (1.22–2.73)	0.009	1.64 (1.14–2.53)	0.03
Severe liver disease (Status 1 or MELD >24)	0.68 (0.50–1.01)	0.07	--	--
Hepatitis C disease	0.72 (0.52–1.02)	0.09	--	--
Packed RBC (per intraoperative unit)	1.00 (1.01–1.04)	0.002	1.52 (0.92–2.22)	0.06
Platelets (per intraoperative unit)	1.01 (1.00–1.03)	0.03	--	--
Roux-en-Y anastomosis*
(1) predicting non-CMV infection within 1 year of LT	2.94 (1.93–4.41)	<0.0001	2.81 (1.91–4.34)	<0.0001
(2) predicting non-CMV infection more than 1 year after LT	0.81 (0.21, 3.52)	0.72	--	--
Ganciclovir as viral prophylaxis	1.64 (1.12–2.44)	0.02	--	--
CMV disease§	3.01 (1.32–7.31)	0.01	2.93 (1.23–6.92)	0.02

NOTE. Variables with P > 0.10 in univariate analysis (UV) and therefore not depicted: age, gender, race, etiology of liver diseases (except HCV as shown); comorbidities; list time; transplant era; baseline: WBC, ANC, creatinine, CKD-EPI eGFR, and albumin; intraoperative cryoprecipitate; ABO-compatible; CMV donor and recipient serogroups; cold ischemia time; type of donor; donor risk index; induction and maintenance immunosuppresion; and ATG induction.

^*Effect of Roux-en-Y anastomosis vs no Roux-en-y anastomosis in predicting non-CMV invasive infection varied over time and was modeled as a time-depentdent covariate to assess its effect (1) the first year of LT and (2) more than 1 year after LT. Roux-en-Y was predictive of non-CMV events within the first year of LT, but not predictive of non-CMV events that occurred more than 1 year of LT.

^§CMV disease, as a time-dependent covariate, was predictive of subsequent non-CMV events. Nonsignificant time-dependent covariates were duration of CMV prohpylaxis, rejection and treatment of rejection with ATG or steroids.

Abbreviations. ANC, absolute neutrophil count; ATG, antithymocyte globulin; CI, confidence interval; CKD-EPI eGFR, Chronic Kidney Disease Epidemiology Collaboration estimated Glomerular Filtration Rate; CMV, cytomegalovirus; HCV, hepatitis C disease; HR, hazards ratio; LT, liver transplantation; MELD, model for end stage liver disease; RBC, red blood cells; UV, univariate; WBC, total white blood cell count.

In multivariable analysis, pretransplant lymphopenia (HR 1.64, CI 1.14–2.53; P = 0.03) and CMV disease (HR 2.93, CI 1.23–6.92; P = 0.02) were significant independent predictors of non-CMV invasive infection in the first 2 years of LT, Table 4. Roux-en-Y anastomosis remained a significant risk factor for predicting non-CMV invasive infection within the first year of LT only (HR 2.81, CI 1.91–4.39; P < 0.0001). Other variables significant in univariate analysis were no longer predictive. Analysis with competing risk of mortality revealed similar results (data not shown).

Lymphopenia subsets

The associated risk between non-CMV invasive infection and pretransplant lymphopenia was explored further when different thresholds for defining lymphopenia were applied: (1) grade 3, severe lymphopenia, ALC ≤ 500 cells/mm³, (2) grade 2, modest lymphopenia, ALC of 501 to 750 cells/mm³, (3) grade 1, mild lymphopenia, ALC 751 to 1000 cells/mm³, and (4) grade 0, no lymphopenia, ALC > 1000 cells/mm³. Kaplan Meier curves to model associated risk of non-CMV invasive infection by different ALC thresholds showed a graded decreased in risk as lymphocyte count increased (log-rank test, P = 0.01), Figure 3A. When the same stratification was modeled for CMV disease, associated risk similarly declines with increasing lymphocyte count but most dramatically from grade 3 to grade 2 lymphopenia (log-rank test, P= 0.0003), Figure 3B.

Figure 3.

Kaplan-Meier estimates of survival free of (A) non-CMV invasive infection and (B) CMV disease in the first 2 years after liver transplantation by grades of lymphopenia: (1) Grade 3, ALC < 500 (mixed dashed line); (2) Grade 2, ALC 501–750 (long dashed line); (3) Grade 1, ALC 751–1000 (short dashed line); Grade 0, ALC >1000 (solid line). Figure 3A, the increased risk of any non-CMV invasive infection is maintained in a stepwise fashion as lymphopenia severity increases, log-rank test P = 0.01. Similarly, in Figure 3B, the more severe the lymphopenia, the lower the probability of surviving without CMV disease, log-rank test P = 0.0003. Abbreviations. ALC, absolute lymphocyte count; CMV, cytomegalovirus.

Death

By one year after LT, 34 of 276 (12%) patients died and by year two, 49 (18%), were deceased. There was no significant difference in the proportion of patients with (21%) versus patients without pretransplant lymphopenia (17%) who died within 2 years, Table 2. However, of the 49 deaths, 8 (16%) had preceding CMV disease, P = 0.04.

Within 2 years of LT, CMV disease (HR 5.62, CI 2.53–12.6; P < 0.001), rejection treated with steroids (HR 1.71, CI 0.97–3.11; P = 0.06), IFI (HR 3.83, CI 1.93–7.34; P < 0.0001), BSI (HR 6.91, CI 3.91–12.2; P < 0.0001) and composite non-CMV infection (HR 6.43, CI 3.53–11.8; P < 0.001) were significant univariate predictors of death. In multivariable analysis, CMV (HR 4.82, CI 3.01–7.82; P < 0.0001), non-CMV invasive infection (HR 6.44, CI 3.51–11.7; P < 0.0001) and rejection treated with steroids (HR 2.22, CI 1.23–4.02; P = 0.007) were the strongest independent predictors of 2-year mortality. Other potential confounding variables including DRI and hepatitis C were not significant in univariate or multivariable analyses (tables not shown).

DISCUSSION

In this retrospective study of primary liver transplant recipients, we demonstrate a strong independent association between pretransplant absolute lymphopenia (ALC ≤ 500 cells/mm³) and the risk of developing both CMV disease and non-CMV invasive infection in the first 2 years after LT. The association between pretransplant lymphopenia and infectious outcomes was independent of known established risk factors for either CMV disease or non-CMV invasive infection.^{1–2, 12–13,17,21,23,30} and suggests an essential contribution of the lymphocyte count as a marker of net immunosuppression after LT.

Our data support the study by Fernandez-Ruiz et al¹² in which they prospectively assessed T cell subpopulations in conjunction with pretransplant total lymphocyte count as potential risk factors for any type of infection after LT in 63 LT recipients. In their univariate analysis, pretransplant CD3⁺ T cell count < 0.75 × 10³/μl, pretransplant CD4⁺ T cell count < 0.5 × 10³/μl, or total lymphocyte counts less than 1.0 × 10³/μl preLT were predictive of infection postLT. In multivariable analysis, T cell subsets were no longer predictive while pretransplant total lymphocyte count remained predictive.

Although Fernandez-Ruiz et al. did not find a significant association with T cell subsets as predictors of any type of postLT infection, more recent studies have explored this idea more rigorously and prospectively and have found evidence to support the use of T cell subpopulations in virologic and immunologic monitoring of CMV to guide preemptive therapy in SOT.^{5–6, 31} These studies do not assess the utility of T cell subsets as markers of risk for non-CMV invasive infection following transplant nor how they can guide clinical decisions. T cell subset counts are not ideal candidate markers of immunosuppression given they are not universally available in all hospital laboratories, have a long processing time, are costly and their relationship to non-CMV infection after SOT is not clear.

We did not examine T cell subsets in the current study because they were not available. We instead demonstrate the broader utility of total lymphocyte counts in immunologic monitoring of both CMV and non-CMV infectious events after LT. Interestingly, the majority of our patients who were lymphopenic were also leukopenic, but no patient was neutropenic prior to LT. Therefore, pretransplant absolute neutrophil count cannot be helpful in predicting infection after LT. We were curious about posttransplant ALC, ANC, and WBC and if there was a relationship between recovery, persistence, and/or new lymphopenia and infection after LT. However, because the median times to non-CMV infection and to CMV disease were less than 30 days and 6 months, respectively, we could only examine counts at 2- and 4-week postLT intervals as potential predictors of our outcomes. In these analyses, only patients with preLT lymphopenia had an elevated risk of any type of postLT infection, regardless of persistent, recovered, or new lymphopenia. It could be that most patients are so immunosuppressed in the first month postLT that no significant difference in immunologic parameters could be measured to assist in risk stratification. Additional prospective studies on these posttransplant hematologic indices at varying time points after LT would help determine if the preLT values are the only ones predictive or if there is a later time point at which the postLT values again become predictive.

Our study has several advantages including: a relatively large cohort, a long follow up period and a large number of potentially confounding variables including several time-dependent covariates. Moreover, we used strict definitions for infectious events and did not include infections such as uncomplicated cystitis and superficial surgical wound infections, where differentiation of true infection from colonization is difficult. We used a more stringent definition of pretransplant lymphopenia of ALC ≤ 500 cells/mm³ rather than < 1000 cells/mm³ based on Bodey’s early work.³ However, even when we stratified infection by different ALC thresholds, the lymphocyte count remained a significant predictor of both non-CMV invasive infection and CMV disease, Figure 3. Importantly, the strength of this associated risk increased according to increased severity of lymphopenia. Perhaps this stepwise effect of lymphopenia severity in predicting infection could be further explored to develop protocols for preemptive therapy akin to opportunistic infection guidelines for directed antimicrobial prophylaxis based on different categories of CD4 count in HIV/AIDS patients.

Despite the strengths of our study, we were limited to a single center, retrospective study, which potentially introduced bias in data collection and event determination. To limit bias, the data was collected while blinded to lymphocyte count and we used previously established definitions of infectious events. Although our models and number of outcomes used conventions of 10 outcomes per variable, due to the small number of CMV events in 2 years of LT, there is the possibility of over-fitting of the data and potential for spurious associations to appear.

To account for potential confounding of decreased lymphocyte counts in patients with decreased renal function,¹³ we excluded dialysis-dependent and combined liver-kidney recipients. Since a slight difference still existed between the groups in the remaining cohort, Table 1, we assessed renal function using several different variables: both continuous (pretransplant creatinine and pretransplant CKD-EPI eGFR) and categorical (pretransplant CKD-EPI eGFR less than 60 and pretransplant CKD-EPI eGFR less than 30). These variables for renal dysfunction were each forced into multivariable models to ensure the relationship between pretransplant lymphopenia and posttransplant infection was independent of any definition of renal dysfunction. If, however, liver-kidney transplant patients were included in the cohort (n=304) and the same analyses performed, the associated risk of pretransplant lymphopenia and posttransplant infection remained robust (data not shown).

In conclusion, pretransplant lymphopenia was an independent predictor of CMV disease and non-CMV invasive infection after liver transplantation. We should use the total lymphocyte count at the time of transplantation as a surrogate marker of native and/or intrinsic immunosuppression to identify at-risk individuals. Once identified, more rigorous, specific surveillance strategies should be applied to preempt infection in high-risk patients. Conversely, less aggressive immunosuppressive protocols may be warranted in certain individuals. We need further prospective studies on the application of this readily available immunosuppressive marker as a predictive tool for all types of infection after any type of solid organ transplantation.

Abbreviations

ALC

absolute lymphocyte count

ANC

absolute neutrophil count

ATG

antithymocyte globulin

BSI

blood stream infection

CAD

coronary artery disease

CKD-EPI eGFR

Chronic Kidney Disease Epidemiology Collaboration estimated Glomerular Filtration rate

CI

confidence interval

CMV

cytomegalovirus

CMV D+

cytomegalovirus seropositive donor

CMV D−

cytomegalovirus seronegative donor

CMVIgG

CMV immunoglobulin

CMV R+

cytomegalovirus seropositive

CMV R−

cytomegalovirus seronegative recipient

D

donor

DRI

donor risk index

EBV

Epstein Barr virus

HIV

human immunodeficiency virus

HCV

hepatitis C liver disease

HR

hazards ratio

IFI

invasive fungal infection

IQR

interquartile range

LT

liver transplantation

MELD

model for end stage liver disease

MMF

mycophenolate mofetil

MV

multivariable

R

recipient

preLT

preliver transplantation

postLT

postliver transplantation

SOT

solid organ transplantation

UV

univariate

vs

versus

WBC

total white blood cell count