Risk Factors for CMV Viremia and Treatment-Associated Adverse Events Among Pediatric Hematopoietic Stem Cell Transplant Recipients.

BACKGROUND: Cytomegalovirus (CMV) causes substantial morbidity and mortality after hematopoietic stem cell transplantation (HSCT). There are limited data on risk factors for CMV viremia and the safety of antiviral medications used to treat CMV in children.
METHODS: We conducted a single-center retrospective study of children who underwent HSCT between 2000 and 2016. We used log-logistic regression to evaluate associations between clinical characteristics and CMV-free survival at 100 days after HSCT. We compared the incidences of laboratory-defined adverse events (AEs) during treatment with ganciclovir and foscarnet.
RESULTS: Among 969 children, the median (interquartile range) age was 6.5 (3.1-11.5) years, and 80% underwent allogeneic HSCT. Two hundred forty-four (25%) children developed CMV viremia. Older age (odds ratio [OR], 0.95; 95% CI, 0.92-0.98), male sex (OR, 0.71; 95% CI, 0.51-0.99), non-Black, non-White race (OR, 0.56; 95% CI, 0.36-0.87), umbilical cord blood donor source (OR, 0.28; 95% CI, 0.08-0.97), and CMV seropositivity (R-/D+: OR, 0.17; 95% CI, 0.07-0.41; R+/D-: OR, 0.14; 95% CI, 0.09-0.21; R+/D+: OR, 0.08; 95% CI, 0.04-0.15) were associated with lower odds of 100-day CMV-free survival. Compared with foscarnet, ganciclovir was associated with lower incidences of thrombocytopenia (incidence rate ratio [IRR], 0.38; 95% CI, 0.15-0.97), electrolyte AEs (IRR, 0.42; 95% CI, 0.24-0.75), endocrine AEs (IRR, 0.52; 95% CI, 0.34-0.79), and renal AEs (IRR, 0.36; 95% CI, 0.19-0.65).
CONCLUSIONS: CMV viremia occurred commonly among children after HSCT, and ganciclovir and foscarnet were associated with distinct toxicity profiles among children with CMV infection. These findings should be considered when developing CMV prevention and treatment strategies for children after HSCT.

Cytomegalovirus (CMV) infection occurs frequently after hematopoietic stem cell transplantation (HSCT) and is associated with substantial morbidity and mortality. In the first year after HSCT, CMV viremia occurs in up to 50% of adult allogeneic HSCT recipients and progresses to end-organ disease in 20% of these patients [1-3]. CMV disease can manifest as pneumonitis, gastritis, colitis, hepatitis, fever, or leukopenia and is associated with a mortality rate as high as 43% in adults [4, 5]. Prior studies in adult HSCT recipients identified risk factors for CMV disease and viremia, including HSCT recipient and donor CMV seropositivity, underlying diagnosis, patient age, conditioning regimen, HSCT donor source, and receipt of immunosuppressive therapies [1, 4, 6–8]. In contrast, there is a significant gap in our understanding of the risk factors for CMV viremia in pediatric HSCT recipients, as prior studies in this population have been limited by small sample size and varied CMV antiviral prophylaxis or surveillance strategies [9-11]. The identification of risk factors for CMV viremia among pediatric HSCT recipients could facilitate the risk stratification of patients for evaluation of new antiviral medications and other novel preventive and treatment strategies for CMV.

Management of CMV in children after HSCT most frequently involves routine surveillance for CMV viremia and initiation of preemptive antiviral therapy. Such an approach has been shown to both limit progression to CMV end-organ disease and improve overall survival among adult HSCT recipients [12]. Ganciclovir and foscarnet are the current agents of choice for the treatment of CMV viremia [13]. These antivirals have similar efficacy but are each associated with significant toxicities; in particular, ganciclovir has frequently been associated with bone marrow suppression while foscarnet has been associated with renal impairment [14-16]. Ganciclovir and foscarnet have been used extensively in infants and children, despite limited safety data. Age-related differences in drug distribution, metabolism, and elimination could affect the safety profiles of these antivirals in children compared with adults [16-18]. Access to safety data is imperative in choosing an antiviral agent for use in a vulnerable population with complex underlying disorders, high risk for CMV end-organ disease, and frequent exposure to other medications.

In this retrospective study, we sought to identify patient and transplant characteristics associated with the risk of CMV viremia in the first 100 days after HSCT in children. As a secondary objective, we compared the incidence of laboratory-defined adverse events (AEs) among children with CMV viremia treated with ganciclovir or foscarnet.

METHODS

Study Design and Population

We conducted a retrospective cohort study of children 1 to 17 years of age who underwent HSCT through the Duke University Pediatric Transplant and Cellular Therapy Program between January 1, 2000, and December 31, 2016. Data were collected from a secure database maintained by the HSCT program, electronic medical records, and the Duke Enterprise Data Unified Content Explorer (DEDUCE) [19]. We limited eligibility to children 1 year of age or older because of the difficulty in interpreting CMV serological testing from younger infants who may still have maternally derived CMV-specific antibodies. We excluded children with missing CMV serologies from either the donor or the recipient (n = 18), children for whom the donor was a CMV-seropositive infant under 1 year of age (n = 3), and children who were receiving foscarnet or ganciclovir on the day of HSCT (n = 19). The study protocol was approved by the Duke Health Institutional Review Board.

Transplant Practices

All HSCT recipients and donors were evaluated for CMV infection before HSCT. Between 2000 and 2005, umbilical cord blood units were not used if maternal serum was positive for CMV immunoglobulin M (IgM); beginning in 2006, units were not used if CMV polymerase chain reaction (PCR) testing of the cord blood unit was positive. Bone marrow donors and transplant recipients were screened for CMV using serum IgM and immunoglobulin G (IgG).

After HSCT, surveillance for CMV viremia among HSCT recipients was performed weekly until day +100 after HSCT. CMV DNA hybrid capture assays on whole blood were used for surveillance from 2000 until 2006; from 2006 onward, weekly quantitative CMV PCR on plasma became standard practice. Antiviral therapy was initiated for any symptomatic CMV infection. Preemptive therapy was initiated after 2 consecutive weeks of CMV PCR >500 IU/mL or after 1 measurement >1000 IU/mL. Children with serological evidence of previous infection with herpes simplex viruses received acyclovir 250 mg/m2/dose given intravenously every 12 hours for the first year after allogeneic HSCT and for 3–6 months after autologous HSCT.

At our institution, children are typically hospitalized until they demonstrate stable donor cell engraftment and are otherwise clinically stable. Most laboratory values used to assess for AEs were measured daily while inpatient and at least weekly during outpatient encounters as part of routine clinical care. Hospital discharge and collection of laboratories were ultimately at the discretion of the clinical provider.

Statistical Analyses

We used multivariable log-logistic regression, a parametric survival analysis, to evaluate associations between patient and transplant characteristics and CMV viremia within the first 100 days after HSCT [20]. Children were censored on the day of death, second HSCT, last recorded health care encounter, or when initiating treatment with an antiviral medication with activity against CMV. We used this model to estimate odds ratios (ORs) of 100-day CMV-free survival for each potential risk factor, with a lower OR indicating a higher odds of CMV viremia by 100 days after HSCT. We similarly used multivariable log-logistic regression to evaluate associations between patient and transplant characteristics and all-cause mortality in the 2 years after HSCT. Multivariable models were adjusted for patient and transplant characteristics, including patient age, sex, race, underlying diagnosis, HSCT type and donor source, conditioning regimen, year of HSCT, and CMV serostatus.

We identified treatment-associated AEs through analysis of laboratory results from patients who developed CMV viremia (Table 1). We used Wilcoxon rank-sum tests to compare baseline laboratory values collected on the first day of therapy or, if unavailable, up to 3 days preceding start of therapy among children starting ganciclovir or foscarnet. We compared incidence rates of AEs during treatment with ganciclovir and foscarnet. Due to the frequency of laboratory abnormalities in this patient population, we did not attribute an AE to receipt of an antiviral if the laboratory abnormality was present in the 3 days before starting antiviral treatment. Moreover, to prevent overestimation of the incidence of treatment-associated AEs, a patient could only meet criteria for a specific AE once during each antiviral treatment course; however, a patient could meet criteria for multiple different AEs during the same treatment course. We lagged laboratory results by 1 day to ensure that AEs were assessed after antiviral exposures. Finally, we used longitudinal, multivariable mixed-effect regression models to compare the effects of ganciclovir and foscarnet treatment on liver enzymes and renal function tests. Laboratory values were log-transformed to achieve normality, and these models were adjusted for clinical factors as above, for day after HSCT as a fixed effect, and for subject as a random effect. Statistical analyses were performed using the survival (version 3.12) and lme4 (version 1.1-26) packages within R, version 4.0.2.

Table 1.

Laboratory-Defined Adverse Events

	AE Value
Electrolyte AEs
Hyponatremia	≤124 mEq/L
Hypernatremia	≥151 mEq/L
Hypokalemia	≤2.5 mEq/L
Hyperkalemia	≥7.0 mEq/L
Bicarbonate: low	≤14 mEq/L
Bicarbonate: high	≥35 mEq/L
Hypocalcemia	≤7.5 mg/dL
Hypercalcemia	≥11.2 mg/dL
Hematological AEs
Leukopenia	≤2 × 103/µL
Leukocytosis	≥100 × 103/µL
Anemia (hemoglobin)	≤7 g/dL
Polycythemia (hemoglobin)	≥24 g/dL
Thrombocytopenia	≤50 × 103/µL
Thrombocytosis	≥1000 × 103/µL
Prothrombin time	≥18 sec
Activated partial thromboplastin time	≥79 sec
Endocrine AEs
Hypoglycemia	≤60 mg/dL
Hyperglycemia	≥200 mg/dL
Gastrointestinal
Aspartate aminotransferase	≥100 U/L
Alanine aminotransferase	≥100 U/L
Alkaline phosphatase	≥500 U/L
Conjugated bilirubin	≥1 mg/dL
Gamma-glutamyl transferase	≥75 U/L
Lipase	≥200 U/L
Renal AEs
Blood urea nitrogen	≥50 mg/dL
Creatinine	≥1.5 mg/dL

Abbreviation: AE, adverse event.

RESULTS

Patient Characteristics

Patient and transplant characteristics of the 969 children included in this study are shown in Table 2. The median (interquartile range [IQR]) age was 6.5 (3.1–11.5) years, and most children were male (61%) and White (69%). Fewer than half (43%) of transplants were performed for a hematological malignancy, and most were allogeneic transplants with an umbilical cord blood donor (59%). Among the allogeneic transplants, 165 (81%) of the bone marrow transplants and 18 (3%) of the umbilical cord blood transplants were from related donors. Few children (5%) received nonmyeloablative or reduced-intensity conditioning regimens. The most common graft-vs-host disease (GVHD) prophylaxis regimens after allogeneic HSCT were a calcineurin inhibitor with mycophenolate mofetil (41%), a calcineurin inhibitor with steroids (39%), and a calcineurin inhibitor with methotrexate (18%). More than half of recipients (58%) and the vast majority of donors (84%) were CMV-seronegative.

Table 2.

Patient and Transplant Characteristics of the Study Population

	No CMV Viremia (n = 725)	CMVViremia (n = 244)	OR	95% CI	P
	No. (%)	No. (%)
Age, median (IQR), y	6.0 (2.8–10.9)	8.5 (4.2–13.3)	0.95	0.92–0.98	.005
Sex
Female	291 (40)	88 (36)	1.00	Ref	-
Male	434 (60)	156 (64)	0.71	0.51–0.99	.04
Race
White	524 (72)	143 (59)	1.00	Ref	-
Black	121 (17)	53 (22)	0.79	0.52–1.20	.27
Other races	80 (11)	48 (20)	0.56	0.36–0.87	.01
Underlying diagnosis
Hematological malignancy	293 (40)	125 (51)	1.00	Ref	-
Genetic or metabolic disorder	139 (19)	38 (16)	0.67	0.41–1.08	.10
Nonmalignant hematological disorder	77 (11)	48 (20)	0.74	0.45–1.21	.23
Immunodeficiency or autoimmune disease	52 (7)	18 (7)	0.78	0.41–1.48	.45
Solid tumor	164 (23)	15 (6)	2.23	0.69–7.24	.18
HSCT type
Autologous	178 (25)	18 (7)	1.00	Ref	-
Allogeneic, bone marrow source	130 (18)	74 (30)	0.46	0.15–1.47	.19
Allogeneic, umbilical cord blood source	417 (58)	152 (62)	0.28	0.08–0.97	.045
Conditioning regimen
Myeloablative	688 (95)	229 (94)	1.00	Ref	-
Nonmyeloablative or reduced intensity	37 (5)	15 (6)	0.72	0.36–1.43	.34
CMV serostatus
R-/D-	482 (66)	57 (23)	1.00	Ref	-
R-/D+	13 (2)	11 (5)	0.17	0.07–0.41	<.0001
R+/D-	143 (20)	128 (52)	0.14	0.09–0.21	<.0001
R+/D+	87 (12)	48 (20)	0.08	0.04–0.15	<.0001

Abbreviations: CMV, cytomegalovirus; D, donor; HSCT, hematopoietic stem cell transplantation; IQR, interquartile range; OR, odds ratio; R, recipient.

Outcomes

Two hundred forty-four (25%) children developed CMV viremia at a median (IQR) of 25 (12–40) days after HSCT. CMV viremia was identified by DNA hybrid capture assay in 120 (49%) children and by PCR in 124 (51%) children. CMV viremia occurred in 226 (29%) children who underwent allogeneic HSCT and 18 (9%) children who underwent autologous HSCT. Of the 539 recipient-seronegative (R-)/donor-seronegative (D-) transplants, there were 53 (10%) primary CMV infections among allogeneic HSCT recipients and 4 (1%) among autologous HSCT recipients. Older age (OR, 0.95; 95% CI, 0.92–0.98), male sex (OR, 0.71; 95% CI, 0.51–0.99), non-Black, non-White race (OR, 0.56; 95% CI, 0.36–0.87), umbilical cord blood donor source (OR, 0.28; 95% CI, 0.08–0.97), and CMV seropositivity in either the donor or recipient (R-/D+: OR, 0.17; 95% CI, 0.07–0.41; R+/D-: OR, 0.14; 95% CI, 0.09–0.21; R+/D+: OR, 0.08; 95% CI, 0.04–0.15) were associated with a lower odds of 100-day CMV-free survival (Table 2). In the 2 years after HSCT, 318 (33%) children died, including 95 (39%) children who developed CMV viremia in the 100 days after HSCT and 233 (32%) who did not develop CMV viremia. CMV viremia was not associated with 2-year all-cause mortality (OR, 1.13; 95% CI, 0.84–1.52).

Treatment-Associated Adverse Events

Of the 244 children who developed CMV viremia, 180 (74%) received antiviral therapy with ganciclovir or foscarnet, including 77 (43%) who received ganciclovir only, 14 (8%) who received foscarnet only, and 89 (49%) who received both ganciclovir and foscarnet. Of the 64 children who did not receive ganciclovir or foscarnet, 4 (6%) received CMV immune globulin, 3 (5%) received valganciclovir, and 1 (2%) received both valganciclovir and CMV immune globulin. The remaining children received no directed treatment for CMV viremia. Children starting foscarnet treatment had lower values of potassium (P = .04), white blood cell count (WBC; P < .0001), platelet count (P = .03), aspartate transaminase (AST; P = .0005), alanine transaminase (ALT; P = .04), alkaline phosphatase (ALP; P = .01), and creatinine (P = .04) than children starting ganciclovir (Table 3). In total, there were 6533 patient-days of ganciclovir monotherapy in 169 treatment courses, 2742 patient-days of foscarnet monotherapy in 103 treatment courses, and 517 patient-days of treatment with both ganciclovir and foscarnet in 78 treatment courses. Foscarnet was started earlier after HSCT than ganciclovir (median [IQR], 34 [14-54] days vs 40 [30-55] days; Wilcoxon-rank sum test, P = .01). Ganciclovir treatment courses were also of longer duration than foscarnet treatment courses (median [IQR], 39.5 [17-64] vs 22 [11–40.5] days; Wilcoxon rank-sum test, P < .0001).

Table 3.

Median Laboratory Values at the Start of Therapy by Antiviral

	Median (IQR) Value		P
	Ganciclovir(n = 169)	Foscarnet(n = 103)
Electrolyte
Sodium, mEq/L	138 (136–139)	139 (136–141)	.24
Potassium, mEq/L	4.1 (3.7–4.5)	4.0 (3.6–4.2)	.04
Bicarbonate, mEq/L	24 (22–26)	24 (22–25.3)	.10
Calcium, mg/dL	9.1 (8.7–9.5)	9.0 (8.6–9.4)	.18
Hematological
WBC, 103/uL	7.9 (4.7–12.25)	1.6 (0.3–7.25)	<.0001
Hemoglobin, g/dL	9.8 (8.8–10.7)	9.6 (8.9–10.5)	.71
Platelet count, 103/µL	49 (31.5–80.5)	38 (24.8–59)	.03
Prothrombin time, sec	12.7 (11.8–13.4)	12.5 (12–13.5)	.89
Activated PTT, sec	29.4 (26.4–35.6)	30.4 (27.1–35.1)	.75
Endocrine
Glucose, mg/dL	114 (91–143)	118 (99–145.2)	.13
Gastrointestinal
AST, U/L	45 (33–66)	36 (25–58)	.0005
ALT, U/L	58 (36–92)	44 (26–89)	.04
ALP, U/L	131 (95–183)	110 (89–147)	.01
Renal
BUN, mg/dL	23 (16–39)	25 (16–45)	.44
Creatinine, mg/dL	0.6 (0.4–0.8)	0.5 (0.3–0.7)	.04

Abbreviations: ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen; IQR, interquartile range; PTT, partial thromboplastin time; WBC, white blood cell count.

We then compared AEs that occurred during ganciclovir treatment, foscarnet treatment, or with no antiviral treatment (Figure 1; Supplementary Table 1). Compared with treatment with foscarnet, ganciclovir treatment was associated with lower incidence rates of thrombocytopenia (incidence rate ratio [IRR], 0.38; 95% CI, 0.15–0.97), electrolyte AEs (IRR, 0.42; 95% CI, 0.24–0.75), endocrine AEs (IRR, 0.52; 95% CI, 0.34–0.79), and renal AEs (IRR, 0.36; 95% CI, 0.19–0.65). No differences in the incidence rates of leukopenia, anemia, and gastrointestinal AEs were observed during treatment with ganciclovir and foscarnet. A significant proportion of the 523 AEs attributed to ganciclovir or foscarnet occurred in the first several weeks of antiviral therapy. Of the 316 AEs attributed to ganciclovir, 79 (25%) occurred during the first 7 days of therapy and 67 (21%) were in the second week of ganciclovir therapy. Of the 207 AEs attributed to foscarnet, 73 (35%) occurring during the first 7 days of therapy and 59 (29%) were in the second week of foscarnet therapy (Supplementary Table 1). Compared with no antiviral therapy, ganciclovir was associated with a 4% (95% CI, 2%–6%) increase in blood urea nitrogen (BUN) per patient-day of therapy, while foscarnet was associated with a 32% (95% CI, 29%–35%) increase in BUN and a 14% (95% CI, 12%–16%) increase in serum creatinine per patient-day of therapy (Table 4, Figure 2). Finally, ganciclovir was associated with a 9% (95% CI, 5%–12%) increase in AST, a 7% increase in ALT (95% CI, 3%–11%), and a 3% increase in ALP (95% CI, 1%–5%) per patient-day of therapy compared with no antiviral therapy, and foscarnet was associated with a 6% decrease in ALT (95% CI, 1%–10%) and a 5% decrease in ALP (95% CI, 7%–2%).

Figure 1.

Incidence of adverse events among children with CMV viremia by antiviral treatment. For the selected AEs, the incidence rates per 1000 patient-days are shown by antiviral treatment during episodes of CMV viremia. Electrolyte AEs include hyponatremia, hypernatremia, hypokalemia, hyperkalemia, low bicarbonate, high bicarbonate, hypocalcemia, and hypercalcemia; endocrine AEs include hypoglycemia and hyperglycemia; GI AEs include aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, elevated conjugated bilirubin, elevated gamma-glutamyl transferase, and elevated lipase; and renal AEs include elevated blood urea nitrogen and elevated creatinine. Error bars represent 95% CIs. Abbreviations: AEs, adverse events; CMV, cytomegalovirus; GI, gastrointestinal.

Table 4.

Association Between Receipt of Antiviral Medications and Changes in Laboratory Values

	β	95% CI
AST
Antiviral
None	Reference
Ganciclovir	1.09	1.05–1.12
Foscarnet	1.01	0.97–1.05
Alt
Antiviral
None	Reference
Ganciclovir	1.07	1.03–1.11
Foscarnet	0.94	0.90–0.99
Alp
Antiviral
None	Reference
Ganciclovir	1.03	1.01–1.05
Foscarnet	0.95	0.93–0.98
Bun
Antiviral
None	Reference
Ganciclovir	1.04	1.02–1.06
Foscarnet	1.32	1.29–1.35
Creatinine
Antiviral
None	Reference
Ganciclovir	1.00	0.98–1.01
Foscarnet	1.14	1.12–1.16

Abbreviations: ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen.

Figure 2.

Trends in laboratory values by day of antiviral therapy with ganciclovir and foscarnet. The changes in selected serum laboratory measures are shown over time by day of therapy with ganciclovir or foscarnet. Points represent the change from the baseline laboratory value at the start of therapy. Fitted lines were estimated by the restricted maximum likelihood, with error shading representing 95% CIs. A, Serum aspartate aminotransferase. B, Serum alanine aminotransferase. C, Serum alkaline phosphatase. D, Blood urea nitrogen. E, Serum creatinine. Abbreviations: ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen.

DISCUSSION

We described the incidence of and risk factors for CMV viremia after HSCT in the largest pediatric cohort studied to date. Without antiviral prophylaxis, CMV viremia occurred in approximately one-quarter of children in the 100 days after HSCT. We found that older age, male sex, non-Black, non-White race, umbilical cord blood transplant, and CMV seropositivity in either the recipient or donor were associated with an increased risk of CMV viremia. Finally, we identified distinct toxicity profiles associated with receipt of ganciclovir and foscarnet in this patient population that could be used to inform treatment decisions for children with CMV viremia after HSCT.

Current estimates of the incidence of CMV viremia in children after HSCT vary widely, from 13% among a Belgian cohort of 65 pediatric allogeneic HSCT recipients to 46% among a Greek cohort of 110 pediatric allogeneic and autologous HSCT recipients [9–11, 21–23]. This wide range is likely an effect of small sample sizes, differing transplantation practices, and geographical differences in CMV seroprevalence [24]. The incidence of CMV viremia in our cohort was within the range reported by prior pediatric studies and slightly lower than incidence estimates of 36%–39% among adult HSCT recipients [5, 7]. The lower incidence in our cohort is likely due to the large proportion of CMV-seronegative umbilical cord blood donors and HSCT recipients who had not yet experienced primary CMV infection, as well as a shorter period of follow-up for our cohort compared with adult studies. While the majority of episodes of CMV viremia in our cohort were likely due to reactivation or donor-derived infection, we did identify a higher cumulative incidence of primary CMV infection in D-/R- transplants (10%) than has been reported in previous pediatric studies (2%–5%) [10, 21, 23]. Notably, patients in our study were not hospitalized for the duration of the study period, suggesting that some of these primary CMV infections may have occurred outside of the hospital environment, and we additionally cannot exclude false-negative serological testing, CMV transmission via blood transfusion products, or nosocomial infections. Although prior studies suggest that CMV reactivation and viremia are associated with increased mortality in adult allogeneic HSCT recipients, we found that all-cause mortality at 2 years after HSCT was similar in children who experienced CMV viremia and those who did not [25, 26]. While this finding may result from the relatively small number of cases of CMV viremia in our cohort, the lack of an association between CMV viremia and mortality suggests that preemptive therapy is a successful strategy for preventing excess mortality related to CMV in children.

CMV seropositivity is an established risk factor for CMV viremia after HSCT. In large retrospective adult studies, R+/D- transplants have been associated with the highest risk for CMV reactivation, followed by R+/D+ transplants [8, 26]. A similar risk hierarchy of donor and recipient seropositivity has not been determined for children; however, CMV seropositivity in the recipient or donor has consistently been a risk factor for CMV infections in pediatric studies [9–11, 21–23]. Our findings confirm the importance of CMV seropositivity in pediatric HSCT recipients while suggesting that R+/D+ serostatus transplants may have a particularly high risk of CMV viremia. Other pediatric studies have reported a high risk of CMV infections among R+/D+ transplants compared with transplants with other serostatus combinations [9, 27], suggesting that donor CMV seropositivity may be less protective against CMV viremia in children than in adults [28]. Taken together, our findings of CMV viremia in children after HSCT can inform optimal CMV serostatus recipient–donor pairings and CMV prophylaxis strategies.

While risk factors for CMV viremia can inform screening and preventive strategies, prior studies evaluating pediatric HSCT patients largely failed to identify associations between patient or transplant characteristics and CMV viremia beyond CMV serostatus [10, 22, 29]. A small cohort study of pediatric HSCT recipients reported more CMV infections among children with older age, leukemia, allogeneic HSCT, exposure to antithymocyte globulin, and acute GVHD; however, in multivariable analyses, these factors were not independently associated with CMV viremia [23]. We identified several risk factors for CMV viremia in our cohort, including patient age, sex, race, and donor cell source. The association between an umbilical cord blood donor and CMV viremia is likely due to the prolonged period of immunosuppression and delayed recovery of CMV-specific T-cell immunity among recipients of these transplants compared with recipients of allogeneic transplants from a bone marrow donor source or autologous HSCT recipients [30]. The effects of patient age may in part be explained by differences in CMV-specific T-cell immunity between young children and adults and the relative importance of donor or recipient T-cell function in controlling CMV viremia after HSCT [6, 31]. Finally, the increased risk of CMV viremia observed with male sex and non-Black minority race could be related to donor selection. In particular, higher degrees of HLA mismatch may necessitate increased use of immunosuppressive medications in these children, which could in turn increase the risk for CMV reactivation [4, 32]. While verifying the mechanisms underlying these risk factors is beyond the scope of the current work, our study identifies several unique factors that may influence CMV risk after HSCT and that warrant further study.

Ganciclovir and foscarnet are the mainstays of CMV therapy after HSCT; however, there is a notable lack of safety data for these antivirals for use in children. Through this retrospective study, we sought to provide real-world experience regarding the safety of these antivirals in children. We observed higher incidences of thrombocytopenia, electrolyte and glucose disturbances, and renal impairment in children treated with foscarnet compared with those treated with ganciclovir. While most of these AEs are consistent with data from adult studies, the frequency of thrombocytopenia events was greater than previously reported in adults [33]. Notably, foscarnet was generally initiated earlier after HSCT than ganciclovir and at a time when graft function likely remained relatively poor. Consistent with a prior report, we did not observe a difference in the rate of leukopenia between ganciclovir and foscarnet, suggesting that leukopenia may be an underrecognized toxicity associated with foscarnet use in children [15]. Finally, there were more renal AEs and larger increases in serum BUN and creatinine over time associated with foscarnet exposure, confirming the substantial nephrotoxicity of foscarnet in children. Importantly, there is recent evidence to suggest that there is a non-dose-dependent association between foscarnet exposure and glomerular filtration rates at 6 and 12 months after HSCT [34]. This potential long-term effect of foscarnet is particularly concerning for children, as renal function progressively declines with age, and an accelerated decline in glomerular filtration rate could lead to early progression to end-stage renal disease [35].

While foscarnet was associated with thrombocytopenia and renal AEs, ganciclovir was associated with gastrointestinal-related toxicity, including elevations of serum AST, ALT, and ALP levels. Similar findings were reported among neonates in a phase II trial of ganciclovir for the treatment of congenital CMV, in which >30% of neonates experienced clinically significant elevations in AST, ALT, and conjugated bilirubin [36]. Interestingly, other studies of ganciclovir use in infants have shown improvement in these laboratory measures with the effective treatment of CMV-associated hepatitis [37, 38]. Evidence of hepatotoxicity in adults receiving ganciclovir is limited to case reports, in which other hepatotoxic medication exposures and concomitant infection may have contributed to laboratory abnormalities [39-41]. Given the lack of evidence of hepatotoxicity associated with receipt of ganciclovir in adult HSCT recipients, our findings suggest that ganciclovir-associated elevations in liver enzymes may be an underappreciated AE of ganciclovir use in pediatric populations and highlight the need for additional safety data for both currently available and new antivirals in pediatric patients.

Our study has several notable limitations. First, we focused on CMV viremia and did not distinguish between CMV infection and CMV end-organ disease. Given that the vast majority of pediatric transplant centers use a preemptive approach for CMV management after HSCT, our focus on CMV viremia is likely to be broadly generalizable to current practice. Second, 2 diagnostic assays were used during different periods of the study. While the level of CMV DNA tends to be higher in whole-blood samples, the PCR assay used in the later years of this study was likely more sensitive in detecting CMV viremia, especially at lower viral loads [42]. This may have led to more CMV viremia diagnoses in later years, though year of transplant was not an independent risk factor for CMV viremia at 100 days. Our analyses also did not account for GVHD diagnosis and treatment or other immunosuppressive therapies. Additionally, we did not have data on neutrophil and lymphocyte counts, and we thus used total WBC as a surrogate for neutropenia and lymphopenia. We also chose to focus statistical modeling on gastrointestinal and renal laboratory values because we anticipated that other laboratory parameters would be impacted by transfusions or other supportive care measures in this population. Finally, inherent to the retrospective study design, patients were not randomized to specific antiviral therapy, and the observed AEs may not be directly related to antiviral exposure.

In conclusion, we found that CMV viremia was common among pediatric HSCT recipients and that children who are older, male, from a non-Black racial minority group, who receive an umbilical cord blood transplant, and who are CMV-seropositive or have a CMV-seropositive donor are at the highest risk of CMV viremia. With our current institutional practices, including preemptive therapy for CMV viremia, children who develop CMV viremia during the first 100 days after HSCT are not at higher risk of all-cause mortality at 2 years after HSCT. We also report that ganciclovir and foscarnet are associated with specific toxicities in children, including hepatotoxicity associated with ganciclovir and leukopenia associated with foscarnet, highlighting the need to consider patient characteristics and baseline organ function in selecting antiviral therapy for CMV infection. Finally, our findings highlight the need for children to be included in clinical trials of antivirals for CMV prophylaxis or treatment to facilitate the development of strategies to reduce the burden of CMV infection in pediatric HSCT recipients.

Click here for additional data file.