Intravenous Acyclovir and Renal Dysfunction in Children: A MatchedCase Control Study

Suchitra Rao, MBBS1, Mark J. Abzug, MD1, Phyllis Carosone-Link, MS, MSPH1, Tori Peterson, PharmD2,

Jason Child, PharmD2, Georgette Siparksy, PhD3, Danielle Soranno, MD4, Melissa A. Cadnapaphornchai, MD4,

and Eric A. F. Sim~oes, MBBS, DCH, MD1,5

Objectives A cluster of children receiving intravenous (IV) acyclovir for meningoencephalitis developed acuterenal failure in April-May 2008, which prompted a retrospective case-control study to determine the rate of andrisk factors for acute nephrotoxicity during IV acyclovir treatment in children.Study design The percentage decrease in glomerular filtration rate in children receiving IV acyclovir who had$1creatinine measurement after acyclovir initiation from October 2006 to January 2009 was classified as renal risk,injury, or failure according to modified Pediatric Risk Injury, Failure, Loss, End-Stage Renal Disease criteria. Univar-iate and multivariate matched analyses were conducted to identify risk factors contributing to nephrotoxicity.Results In the selected study group, renal dysfunctionwas seen in 131 of 373 (35%) treatment courses studied: 81of 373 (22%) risk, 36 of 373 (9.7%) injury, and 14 of 373 (3.8%) failure. Most renal dysfunction occurred within48 hours of the initiation of acyclovir. Renal function returned to the normal range but not to baseline in most casesduring the follow-up period. Risk factors for renal dysfunction included acyclovir dose >15 mg/kg (OR 3.81, 95%CI1.55-9.37) for risk; cumulative exposure greater than calculated cumulative exposure based on 500 mg/m2/dose(OR 6.00, 95% CI 1.95-18.46) for injury; and age >8 years (OR 21.5, 95% CI 2.2, >1000) and ceftriaxone coadmin-istration (OR 19.3, 95% CI 1.8, >1000) for failure.Conclusions Nephrotoxicity associated with IV acyclovir is common and necessitates renal function monitoring.Risk factors include greater dose, older age, and concomitant ceftriaxone administration. Outside the neonatalperiod, renal dysfunction may beminimized by dosing IV acyclovir below thresholds associated with nephrotoxicity(ie, #500 mg/m2/dose or #15 mg/kg/dose), particularly in older patients. (J Pediatr 2015;166:1462-8).

See editorial, p 1341

cyclovir is used commonly in children for the empiric treatment of meningoencephalitis and neonatal sepsis and for

A treatment of suspected or proven herpes simplex virus (HSV) and varicella zoster virus (VZV) infections. Recommen-ded dosing regimens vary depending on the offending virus (HSV or VZV) and clinical indication. Dosing acyclovir at

20mg/kg by intravenous (IV) every 8 hours is recommended for neonatal HSV infections. The Food andDrug Administration–approved dose for HSV encephalitis in children 3 months of age to 12 years is also 20 mg/kg every 8 hours, but many expertsrecommend 30-45 mg/kg/day for this age group. For children 12 years of age and older, 10 mg/kg every 8 hours is recommen-ded for HSV encephalitis, with consideration by some sources of dosing based on ideal body weight.1 For VZV infections, rec-ommended doses include 10 mg/kg IV every 8 hours or 500 mg/m2 every 8 hours.1-3

Studies in animals and humans demonstrate that acyclovir can cause nephrotoxicity characterized by intrarenal obstructivenephropathy secondary to drug crystal formation in the collecting ducts.4,5 Interstitial nephritis and tubular necrosis also canresult in renal insufficiency, and it is hypothesized that direct tubular injury causes rapid increases in serum creatinine (sCr)levels within 12 hours of initiation of treatment.6,7

From April-May 2008, we observed 4 cases of acute renal failure in patients with meningoencephalitis receiving 20 mg/kg/dose of IV acyclovir every 8 hours. In all cases, patients concomitantly received vancomycin, ceftriaxone, and IV contrast for

imaging studies and developed acute nephrotoxicity within 24-72 hours of initi-

From the Departments of 1Pediatrics (InfectiousDiseases and Hospital Medicine) and 2Pharmacy,University of Colorado School of Medicine andChildren’sHospital Colorado; 3Department of Clinical Informatics,Children’s Hospital of Colorado; 4Section of Nephrology,University of Colorado School of Medicine andChildren’sHospital Colorado; and 5Department of Epidemiology,Colorado School of Public Health and Children’s HospitalColorado, Aurora, CO

The authors declare no conflicts of interest.

0022-3476/$ - see front matter. Published by Elsevier Inc.

http://dx.doi.org/10.1016/j.jpeds.2015.01.023

BMI Body mass index

eGFR Estimated glomerular filtration rate

HSV Herpes simplex virus

IV Intravenous

sCr Serum creatinine

VZV Varicella zoster virus

Ddose500 Difference between total cumulative acyclovir dose before attaining worst renal

dysfunction category and a calculated (potential) cumulative exposure based on

receipt of 500 mg/m2/dose

1462

Vol. 166, No. 6 � June 2015

ating acyclovir therapy. These observations prompted ourstudy of the rate of and potentiating risk factors for acuterenal failure in children receiving IV acyclovir.

Methods

We conducted a retrospective case-control study of patientsadmitted to Children’s Hospital Colorado from October2006 to January 2009 who received treatment doses of IVacyclovir. Approval was obtained from the Colorado Multi-ple Institutional Review Board. Patients with $1 sCr mea-surement after initiation of IV acyclovir were included.Hematology/oncology patients and those receiving oralacyclovir were excluded.

The Schwartz equation was used to calculate baseline esti-mated glomerular filtration rate (eGFR) before the administra-tion of acyclovir, based on sCr level and height adjusted forbody surface area.8 For patients lacking a baseline sCrmeasure-ment, the normal mean sCr for age was used.8 In patients forwhom height was not documented, height was derived fromweight-for-length curves using the same percentile as the pa-tient’s weight for age. Subsequent eGFRs were determined onthe basis of each sCr measurement after acyclovir was started,and the percentage decrease in eGFR from baseline was calcu-lated. For patients with a baseline sCr before the initiation ofacyclovir and follow-up sCr values available, the final percent-age change in eGFR was calculated using the last available sCrfor each treatment course. Patients developing renal dysfunc-tion >1 week after the end of acyclovir administration wereexcluded, given the possibility of other causes of nephrotoxi-city. Acute kidney injury was graded according to changes inestimated creatinine clearance as defined by the PediatricRisk Injury, Failure, Loss, End-Stage Renal Disease classifica-tion system.9 Decreases in eGFR of 25%-49% defined risk;50%-74% defined injury, and$75% defined failure. The firsteGFR decrease that fulfilled the criterion for the patient’s worstcategory of dysfunction attained during acyclovir treatmentwas included for analysis.

Caseswere defined as patients who developed renal dysfunc-tion (risk, injury, or failure). For each case, we identified allpossible controls without renal dysfunction that had receivedat least the same number of IV acyclovir doses as the case atthe time of the patient’s worst category of renal dysfunction.We used random sampling among this pool to identifymatched controls. Controls were used only once for matchingwithin each renal dysfunction stratum (risk, injury, failure);however, because the pool of controls was constant, the samecontrol couldbematched to cases indifferent renal dysfunctiongroups. On the basis of the frequency of various levels of renaldysfunction,wematched controls in the following case: controlratios: risk, 1:2; injury, 1:4; and failure, 1:5.

Statistical AnalysesThe primary outcome was the rate of renal risk, injury, andfailure among IV acyclovir recipients. Secondary outcomemeasures were admission to the intensive care unit, dialysisrequirements, and length of hospitalization. Risk factors

analyzed included age, sex, race, weight, body mass index(BMI), acyclovir dose and duration, and concomitant neph-rotoxic antimicrobials and contrast agents.Descriptive statistics were used to report demographic and

clinical characteristics. For aggregate comparisons of cases andcontrols for each of the risk, injury, and failure groups, aMantel-Haenszel c2 test was computed for categorical vari-ables. For continuous variables with normal distributions,mean values for case and control groups were comparedwith the independent samples t test, and for those with non-normal distributions, median values were compared usingthe Mann-Whitney U test (SPSS, Inc, v.19; IBM, Armonk,New York). P values <.05 were considered significant.Matched univariate analyses for the risk, injury, and failure

groups were performed with dichotomous variables. Forcontinuous variables, distributions for cases and controlswere graphed. For those distributions that were sufficientlydisparate, receiver operating characteristic curves weregenerated to model sensitivity and specificity for each vari-able in distinguishing cases from controls (Figure 1;available at www.jpeds.com). Values that optimizedsensitivity and specificity were chosen as cut points andused for subsequent dichotomous comparisons in matchedanalyses.To assess the relationship between acyclovir dosing and

nephrotoxicity (risk, injury, and failure groups), we per-formed 2 evaluations. Initially, univariate matched analyseswere conducted by comparing acyclovir doses of #15 mg/kg/dose vs >15 mg/kg/dose and #500 mg/m2/dosevs >500 mg/m2/dose, for each outcome. The second methodcompared the total cumulative amount of acyclovir adminis-tered to each subject before attaining his/her worst renaldysfunction category to a calculated (potential) cumulativeexposure based on receipt of 500 mg/m2/dose. Differencesbetween the 2 exposures (Ddose500) >0 reflect actual cumu-lative exposures that are greater than expected based on a500 mg/m2/dose. For controls, the actual exposure was trun-cated at the same number of acyclovir doses as its matchingcase to calculate the Ddose500.For comparing outcomes in cases and controls, matched

univariate analyses were performed using Epi Info V3.5.3(Centers for Disease Control and Prevention, Atlanta, Geor-gia), yielding ORs and 95% CIs. Variables with a P# .2 wereincluded in subsequent multivariate conditional logisticregression analyses for the risk and injury groups to deriveORs and 95% CIs for model predictors. The PHREG proce-dure (SAS version 9.3; SAS Institute Inc, Cary, North Car-olina) was performed with the model parameters that wereidentified as significant in the univariate analyses. A stratumwas used for each matched set, and FORWARD selection wasprespecified.10 For interrelated variables (eg, body surfacearea, weight, age, and dosing variables), only the most signif-icant dose variable was used in the model. Because patientswho received antibiotic therapy in conjunction with acyclovirgenerally received either ceftriaxone or cefotaxime, only cef-triaxone was used in multivariate analyses. For the failuregroup, a multivariate analysis for sparse data, the Conditional

1463

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 166, No. 6

Exact Logistic Regression test (SAS Proc Logistic) was used.11

Predictors with P < .001 in the univariate analysis that werenot highly correlated with age were entered into the model.

To further explore acyclovir dosing using weight and sur-face area-based regimens, representative plots of dose forweight using a variety of dosing regimens (250 and 500 mg/m2 per dose and 10, 15, and 20 mg/kg per dose) were gener-ated. The 50th percentile weight-for-length was used forbody surface area calculations in these representative plots.

Results

IV acyclovir was administered to 782 patients over 813 hos-pitalizations. At least one sCr measurement was obtained in537 patients over 556 hospitalizations. Six hospital encoun-ters were excluded because of the development of renaldysfunction >1 week after the end of acyclovir administra-tion. Of the 550 remaining hospitalizations, an sCr wasobtained after initiation of acyclovir in 373. These 373 hospi-talizations (371 patients; 2 patients had 2 hospitalizationsthat included IV acyclovir treatment) comprise the studygroup. An sCr before acyclovir was obtained in 250 of these373 hospitalizations. Height was not available for 11.2% ofpatients and was extrapolated from weight-for-length curves.There was no difference in baseline eGFR and first eGFRdecrease qualifying for worst renal category between patientswith calculated heights vs those with heights available (P = .1and P = .1, respectively). The baseline eGFR for all patients inthe study group were in the normal range. The age distribu-tions and ceftriaxone, cefotaxime, and vancomycin usageamong cases and controls were similar (Figure 2 andTable I; available at www.jpeds.com). Conditions treatedwith IV acyclovir were non-HSV/VZV meningoencephalitis(40%), non-HSV/VZV sepsis/viral syndrome (28%), andnon-HSV/VZV neonatal illness (23%), in addition to 7%proven HSV and 1% proven VZV infections, with a similardistribution among cases and controls. Initial indicationsassociated with IV acyclovir therapy are shown in Figure 3(available at www.jpeds.com).

In the selected study group of acyclovir recipients with atleast 1 creatinine measurement after acyclovir initiation,renal dysfunction occurred in 131 of 373 (35%) treatmentcourses studied: 81 of 373 (22%) risk, 36 of 373 (9.7%)injury, and 14 of 373 (3.8%) failure (Table II). Mediantimes to risk, injury, and failure were 0.8, 0.7, and 1.4 days,respectively. In aggregate analyses, failure cases were older,had greater BMI and weight, and received fewer doses ofacyclovir compared with failure controls (Tables II and IIIand Figure 4; Table III and Figure 4 available at www.jpeds.com). Pre-existing renal conditions were identified in2 controls (both in the injury group with normal baselineeGFRs) and in no cases (P = .9, injury cases vs controls).Urinalysis abnormalities were seen in 18 of 42 (43%), 14 of23 (61%), and 4 of 9 (44%) cases with results available inthe risk, injury, and failure groups, respectively, comparedwith 45 of 113 (40%) of controls. Hematuria, proteinuria,white blood cells, and granular casts were the most

1464

common findings in patients with renal dysfunction, butwithout significant differences between cases and controls.No patients required dialysis. Although sCr returned to thenormal range after acyclovir cessation or dose reduction forall cases who had serial sCr available, eGFR did not returnto baseline in the majority by the end of follow-up(Table II and Figure 5; Figure 5 available at www.jpeds.com). There were no significant differences between casesand controls in admission to the intensive care unit,duration of acyclovir treatment, or length of hospitalization.Univariate matched analyses demonstrated that: age

>8 years, weight >20 kg, and BMI >19 kg/m2 were risk factorsfor development of renal failure; acyclovir dose >15 mg/kgwas associated with renal risk; and dose >500 mg/m2 andDdose500 >0 were associated with nephrotoxicity in all cate-gories (Table IV). Medication risk factors for renal failureincluded concurrent administration of ceftriaxone,gadolinium, and their concomitant use. Associations withother antibiotics, contrast agents, and their combinationswere not significant.Multivariate conditional logistic regression revealed that

acyclovir dosing >15 mg/kg was a significant predictor forrenal risk (Table V). Ddose500 >0 was associated with renalinjury. Age >8 years and ceftriaxone usage were significantpredictors of renal failure, despite relatively small numbersof subjects with failure. Because data in the failure groupwere sparse, we were unable to examine the interactionbetween age >8 years and ceftriaxone in the multivariatemodel. Additional sensitivity analysis limited to patientsgreater than 1 year of age demonstrated the samesignificant predictors. Ceftriaxone use was more frequent infailure cases and controls >8 years of age, although it wasnot used exclusively in older children (12/64 cases andcontrols <8 years vs 11/20 >8 years, P = .0015). Amongcases and controls >8 years of age, ceftriaxone use wasmore frequent among failure cases (8/10) than controls(3/10) (Table V and Figure 6; Figure 6 available at www.jpeds.com).Figure 7 demonstrates the acyclovir dose a patient would

receive as a function of weight using different dosingregimens. Beyond a weight of 20 kg, the acyclovir doseusing 10 mg/kg/dose would exceed the dose using 250 mg/m2/dose, and the dose using 20 mg/kg/dose would exceedthe dose using 500 mg/m2/dose. Beyond 51 kg, the doseusing 15 mg/kg/dose would also exceed the dose using500 mg/m2/dose.

Discussion

This study demonstrates that nephrotoxicity in childrenreceiving IV acyclovir is common, with renal dysfunctionnoted in 35% of acyclovir recipients with at least one creati-nine measurement after acyclovir initiation who comprisedthe study group. The rate of development of acute renal fail-ure (as defined by Pediatric Risk Injury, Failure, Loss, End-Stage Renal Disease injury and failure groups) was 13.4%,in concordance with previous reports.6 Median times to renal

Rao et al

Table II. Patient characteristics, hospital course, and change in eGFR from baseline

Patients Normal, n = 240

Risk Injury Failure

Case, n = 81 Control,* n = 160 Case, n = 36 Control,* n = 143 Case, n = 14 Control,* n = 70

Male, n (%) 144 (60) 43 (53.1) 102 (63) 20 (55.6) 82 (57.3) 6 (42.9) 43 (61.4)Age in weeks,median (range)

5 (0-1003) 8 (0-995) 8.5 (0-1003) 7.5 (0-892) 5 (0-1003) 747 (1-955) 7 (0-962)†

Race, n (%)White 113 (47.1) 34 (42) 65 (41.3) 18 (50) 66 (46.2) 3 (21.4) 27 (40.3)Black 17 (7.1) 3 (3.7) 12 (7.5) 0 (0) 6 (4.2) 2 (14.3) 5 (7.5)Hispanic 52 (21.7) 18 (22.2) 37 (23.8) 11 (30.6) 31 (21.7) 3 (21.4) 12 (17.9)

BMI, median (range) 14.3 (14.4-5.0) 14.5 (10.2-29.8) 14.4 (4.9-37.4) 14.2 (10.2-26.3) 14.3 (8.4-151.5) 21.4 (11-31) 14.2 (8.9-31.6)†

BMI percentile,median (range)

27.4 (0.1-99.9) 32.3 (0.1-98.6) 32.1 (0.1-99.9) 34.8 (0.1-99.9) 31.6 (0.1-99.9) 56.3 (0.7-99.9) 25.3 (0.1-98.4)

Weight in kg,median (range)

4.4 (1.5-111) 4.9 (1.5-111.2) 4.9 (1.5-111.2) 5.5 (1.9-60.1) 4.56 (1.5-99.4) 50.3 (2.7-88.1) 5.1 (2-99.4)†

Weight percentile,median (range)

26.2 (0.1-99.9) 30.6 (0.1-99.3) 31.2 (0.1-99.9) 16.2 (0.1-99.9) 30.0 (0.1-99.9) 45 (1.1-99.3) 42.1 (0.1-99.9)

Hospital encounters n = 242 Case, n = 81 Control,* n = 162 Case, n = 36 Control,* n = 144 Case, n = 14 Control,* n = 70

Days to worst categoryof renal impairment,median (range)

- 0.8 (0-14.4) - 0.7 (0-13.0) - 1.4 (0.4-4.4) -

ICU admission, n (%) 100 (41.7) 37 (45.1) 64 (40.0) 36 (50) 66 (46.1) 8 (57.1) 34 (48.6)Number of doses ofacyclovir during hospitalization,median (range)

7 (1-63) 7.5 (1-64) 8 (2-63) 8 (1-65) 8 (1-63) 6.5 (2-66) 8 (1-60)†

Number of days on acyclovir,median (range)

2.3 (0.3-21.0) 2.3 (0.3-21.4) 2.3 (0.3-21.0) 2.6 (0.3-21.3) 2.3 (0.3-21.0) 2.3 (0.3-22) 2.7 (0.3-56.0)

Length of stay, median (range) 7 (1-160) 6 (1-80) 7 (1-140) 16 (1-94) 9 (1-160) 8 (4-35) 9 (2-160)

Acyclovir courses withbaseline sCr

Case, n = 41 Control, n = 117 Case, n = 21 Control, n = 111 Case, n = 6 Control, n = 36

Median % change ineGFR from baselinex

�0.33 25.88† �40 35.44† �33.45 33.93z

Return of eGFR to baseline, n (%){ 12 (29.27) 74 (63.25)† 7 (33.33) 86 (77.48)† 1 (16.7) 30 (83.3)†

Median number of sCr afterstart of ACV (range)

4 (1-21) 2 (1-30) 9 (1-26) 3 (1-30) 7.5 (1-20) 2 (1-20)

Median number of days afterbaseline follow-up sCravailable (range)

3 (0-42) 4 (0-31) 13 (0-48) 4 (0-30) 7 (0-22) 3 (0-28)

ICU, intensive care unit; ACV, acyclovir.*Controls within each group are independent subjects; however, a control may be used in more than 1 group.†P # .01.zP # .002.xMedian percentage change in eGFR from baseline is based on last sCr available for each subject. Negative values reflect a decrease in eGFR, and positive values reflect an increase in eGFR frombaseline. Data are limited to patients with baseline sCr available before the start of acyclovir.{Number of patients with return to baseline eGFR is based on last sCr available for each subject. Return to baseline is defined as subject’s last percentage change in eGFR being greater than or equalto 0, indicating that it is equivalent to or exceeds that subject’s baseline eGFR.

June 2015 ORIGINAL ARTICLES

dysfunction were within 48 hours of initiation of acyclovir. Inunivariate analyses, risk factors for renal failure included age>8 years; weight >20 kg; BMI >19 kg/m2; and use of ceftriax-one, gadolinium, or its combination. For all categories ofrenal dysfunction, >500 mg/m2 of acyclovir/dose andDdose500 >0 were significant risk factors. Multivariate anal-ysis demonstrated that risk factors for renal dysfunctionincluded older age, greater acyclovir dose, and concomitantadministration of ceftriaxone.

Numerous reports, mostly of adults, describe renaldysfunction with high-dose IV acyclovir. The majority ofpatients developed acute renal failure within 1-3 days, oftenwith crystalluria on polarized microscopy. Renal functiongenerally returned to normal after cessation or dose-reduction of acyclovir.6 Our study demonstrated similar ratesand time to development of nephrotoxicity in children, withsimilar reversibility. Nephrotoxicity disproportionately

Intravenous Acyclovir and Renal Dysfunction in Children: A Matc

affected older and larger children and those who receivedgreater dosing of acyclovir, whether examined as a functionof single dose or cumulative exposure. Larger patients whoreceive greater doses may be at greater risk of crystalnephropathy, which occurs when the collecting duct concen-tration of acyclovir exceeds drug solubility. Infants andyoung children may be less susceptible to acyclovir nephro-toxicity because of diminished renal concentrating ability,protecting them from intratubular crystallization. Also, olderpatients may be more likely to receive other nephrotoxicmedications not examined in this study.12

The increased potential for acyclovir-associated nephrotox-icity when combined with other nephrotoxic agents has beendescribed.13,14 We show a specific association with the coad-ministration of ceftriaxone, but not other potentially nephro-toxic antimicrobials such as vancomycin and gentamicin,perhaps because acyclovir and ceftriaxone share similar

hed Case Control Study 1465

Table IV. Univariate matched analysis

Risk Injury Failure

OR (95% CI) P value OR (95% CI) P value OR (95% CI) P value

DemographicsMale 0.66 (0.38-1.14) .10 0.95 (0.48-1.89) .75 0.52 (0.18-1.53) .11Race

White/nonwhite 0.98 (0.58-1.65) .82 1.14 (0.56-2.34) .85 0.43 (0.11-1.71) .15Black/nonblack 0.50 (0.14-1.77) .17 - - 2.25 (0.37-13.51) .73Hispanic/non-Hispanic 0.97 (0.52-1.81) .79 1.59 (0.72-3.56) .37 1.38 (0.30-6.28) 1.00

Age >8 y* 0.77 (0.37-1.58) .36 1.71 (0.70-4.17) .33 41.00 (3.78-444.57) <.001Weight >20 kg* 0.78 (0.38-1.57) .38 1.76 (0.72-4.31) .31 20.50 (2.79-150.39) <.001BMI >19 kg/m2* 1.12 (0.50-2.48) .94 1.14 (0.37-3.49) .95 20.00 (2.87-139.39) <.001

Acyclovir doseDdose500 >0

† 2.89 (1.21-6.88) .02 6.20 (1.99-19.35) <.001 19.00 (3.77-95.81) <.001Dose >500 mg/m2 2.89 (1.21-6.88) .02 6.20 (1.99-19.35) <.001 12.67 (3.18-50.45) <.001Dose >15 mg/kg 4.38 (1.65-11.57) <.01 2.00 (0.70-5.73) .27 0.56 (0.15-2.13) .23

Concomitant drugsCefotaxime 1.20 (0.70-2.07) .61 0.29 (0.11-0.76) <.01 0.35 (0.08-1.52) .07Ceftriaxone 2.07 (0.98-4.37) .09 1.88 (0.80-4.45) .21 38.00 (3.47-415.85) <.001Gentamicin 0.71 (0.31-1.63) .32 0.95 (0.37-2.46) .73 0.50 (0.10-2.56) .24Vancomycin 1.64 (0.68-3.97) .40 1.67 (0.57-4.85) .52 - -Ioversol 2.00 (0.40-9.91) .67 4.00 (0.81-19.82) .18 - -Gadolinium 1.08 (0.43-2.70) .94 1.70 (0.53-5.42) .57 4.75 (1.34-21.22) .03Ceftriaxone + gadoliniumz 2.67 (0.60-11.92) .35 3.20 (0.83-12.40) .14 10.00 (1.44-69.26) <.01

ROC, receiver operator characteristic.*ROC curves were generated for continuous variables to model sensitivity and specificity for each variable in distinguishing cases from controls. Cut points, which optimized sensitivity and specificity,were used for subsequent dichotomous comparisons in matched analyses (Figure 1).†Actual cumulative exposure of acyclovir to the time of the case’s worst category of renal dysfunction greater than the calculated cumulative exposure based on administering 500 mg/m2/dose.zOther drug and contrast combinations were not statistically significant.

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 166, No. 6

mechanisms of injury. An additive association of ceftriaxonehas been suggested in previous reports in adults. Vomieroet al15 described 3 adult patients treated with ceftriaxone andacyclovir who developed nonoliguric renal failure within2-3 days. Renal biopsy and urinary low-molecular-weight pro-teins analysis suggested a direct tubulopathy. Urolithiasis is anuncommon complication of ceftriaxone in children and adults.Intratubular crystallization of ceftriaxone could be additive totubular obstruction resulting from acyclovir crystalliza-tion.16,17 Acute interstitial nephritis, reported to occur at anincreased rate when ceftriaxone is coadministered with vanco-mycin, may further contribute to enhanced nephrotoxicityseen with the combination of ceftriaxone and acyclovir.18

Our exact test suggested that ceftriaxone and age are indepen-dent predictors of renal failure, although small numbers in thisanalysis precluded statistically ruling out an interaction oradjusting for confounding variables. The reason for ceftriaxoneemerging as a risk factor only in the renal failure group isunclear. We speculate that this subgroup reached a criticalthreshold of renal dysfunction that rendered them vulnerableto additive effects of ceftriaxone. It is also possible that ceftriax-one use was a marker for more severe acute illness.

Coadministration of gadolinium, which can induce acutetubular epithelial injury,19 was also a risk factor for renal fail-ure in the univariate analysis, but, perhaps because of smallsample size, was not significant in the multivariate analysis.We did not find incremental nephrotoxicity with iodine-containing contrast, despite similar nephrotoxic potential.20

This may reflect the small number of patients who receivedioversol and that computed tomography scans are often theinitial radiologic study performed in children with meningo-

1466

encephalitis, before the administration of acyclovir. Magneticresonance imaging studies with gadolinium aremore likely tobe performed later in the disease course while patients receiveacyclovir.Our study affirmed that acyclovir is commonly prescribed

as empiric therapy with relatively few cases of actual HSVand VZV disease. However, multidrug regimens that includeacyclovir may not be benign. Acyclovir-induced nephrotox-icity is common and may complicate clinical care. It maynecessitate adjustments in fluid administration and dosingof renally excreted medications. Symptoms of acyclovirtoxicity (which may be accentuated when drug levels in-crease in the setting of associated nephrotoxicity) includemalaise, nausea, headache, irritability, tremulousness, andconfusion,21,22 which can mimic symptoms of meningoen-cephalitis and exacerbate patients’ altered neurologic states.In patients with severe renal dysfunction, symptoms of ure-mia may additionally confuse the neurologic picture.Finally, acyclovir-associated nephrotoxicity can have majoracute and chronic impact on renal function. Although nopatients in our study required dialysis, requirement for renalsupport and irreversible renal failure have been reported inadults with acyclovir-associated nephrotoxicity.23-25 ThesCrs among cases with serial measurements in our study re-turned to the normal range but did not return to baselinevalues in the majority of cases. There was a wide range inthe number of sCrs obtained among cases and controlsand lack of standardized follow-up, and we do not knowwhether eGFRs would have reached baseline with longerfollow-up. Therefore, the clinical relevance of this observa-tion is unknown.

Rao et al

Table V. Multivariate conditional logistic regression ofrisk factors for renal dysfunction

Category Model predictorOR (95% CI) for

significant predictorsTest statistic

P value

Risk Acyclovir dose >15 mg/kg 3.81 (1.55-9.37) .002*Injury Ddose500

† >0 6.00 (1.95-18.47) .0004*Failurez Age >8 y 21.5 (2.2, >1000) .0033x

Ceftriaxone 19.3 (1.8, >1000) .0063x

Variables with P # .2 in the univariate matched analyses (Table IV) were included in theindividual risk and injury models. For the failure model, variables with P < .001 in theunivariate analysis that were not highly correlated with age were included. For acyclovirdose variables, the most significant dose variable was entered into each model. Variablesincluded for analysis were: risk - male, black race, acyclovir dose >15 mg/kg, ceftriaxone;injury - Ddose500, ceftriaxone, gadolinium, gadolinium + ceftriaxone (ioversol was notincluded because of the small sample size, whereas gadolinium was included despite aunivariate analysis P > .2 because the model included the interaction term ofgadolinium + ceftriaxone); failure - age >8 y and ceftriaxone (dose >500 mg/m2 and weight>20 kg were not included because they were highly correlated with age, [R = 0.52 andR = 0.97, respectively, P # .0001]).*Wald c2 P value from conditional logistic regression using a maximum likelihood ratio algo-rithm.†Actual cumulative exposure of acyclovir to the time of the case’s worst category of renaldysfunction greater than the calculated cumulative exposure based on administering500 mg/m2/dose.zThe interaction term destabilized the model, so we could not statistically test the interaction ofage with ceftriaxone.xConditional exact logistic regression test used to derive P value and OR estimates.

June 2015 ORIGINAL ARTICLES

A strength of this study lies in the matching methodsused. Individual matching of cases and controls was per-formed to ensure that controls had received at least thesame number of doses of acyclovir as cases as of the timecases reached their worst renal category. Age-matchingwas not undertaken, given that this was a potentially impor-tant variable to be considered in the risk factor analysis.More than one control was matched per case to increasethe power of the study in light of the small numbers of pa-tients in the injury and failure groups.

Figure 7. Amount of acyclovir in milligrams a patient would receivAs the intersecting lines demonstrate, beyond a weight of 20 kgexceed the dose using 250 mg/m2/dose, and the dose using 20 mBeyond 51 kg (right arrow), the dose of acyclovir using 15 mg/kg

Intravenous Acyclovir and Renal Dysfunction in Children: A Matc

This study has several limitations. It was not designed toprove causality of renal dysfunction by acyclovir but insteadto explore the incidence and risk factors for this association.A relationship between acyclovir and renal dysfunction wasassumed based on previous literature reports, the temporalassociation of acyclovir administration and worsening renalfunction, and improvement in renal function after dosereduction or cessation. It is important to note that the studypopulation represented the subset of patients receivingacyclovir that had at least 1 creatinine measurementobtained after IV acyclovir initiation, which may haveskewed our population to those with a more severe illnesspresentation, who were receiving multiple potentially neph-rotoxic agents, or for whom there was clinical concern fornephrotoxicity. Many patients in the study did not havesCrs measured before the administration of acyclovir andwere assumed to have normal renal function at baseline.Other studies similarly make this assumption, becauseincluding only patients with baseline values would reducethe sample size significantly.Height measurements were not available for 11% of

patients. For these patients, height was determined fromweight-for-length curves on the basis of their weight percen-tiles; there were no significant differences in baseline orchange in eGFR between subjects with inferred height valuesvs those with height measurements. Finally, other risk factorspotentially implicated in acyclovir-associated nephrotoxicity,including volume depletion, concentration and infusion rateof acyclovir, volume of concomitant fluids, illness severityand other nephrotoxic agents, such as nonsteroidal anti-inflammatory agents, were not investigated.4,5,26

In conclusion, renal impairment in children receiving IVacyclovir is frequent and occurs within several days of drug

e for his/her weight using different acyclovir dosing regimens.(left arrow), the dose of acyclovir using 10 mg/kg/dose wouldg/kg/dose would exceed the dose using 500 mg/m2/dose.

/dose would exceed the dose using 500 mg/m2/dose.

hed Case Control Study 1467

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 166, No. 6

initiation. Adherence to guidelines for dosing, monitoring ofrenal function at baseline and from early in the treatmentcourse, and adequate hydration status are important consid-erations when administering IV acyclovir. Renal functiongenerally improves after dose reduction or cessation. Renaldysfunction appears to be dose-dependent and occurs morecommonly with older age and concomitant administrationof ceftriaxone. Outside the neonatal period, renal dysfunc-tion may be minimized by dosing IV acyclovir below thresh-olds we found to be associated with nephrotoxicity, ie,#500mg/m2/dose or#15mg/kg/dose (and perhaps employ-ing body surface area-based dosing in larger patients). n

Submitted for publication Apr 29, 2014; last revision received Dec 15, 2014;

accepted Jan 13, 2015.

Reprint requests: Suchitra Rao, MBBS, Department of Pediatrics (Infectious

Diseases), University of Colorado School of Medicine and Children’s Hospital

Colorado, Box 055, 13123 E. 16th Ave, Aurora, CO 80045. E-mail: suchitra.

rao@childrenscolorado.org

References

1. Bradley JS, Nelson JD.Nelson’s pediatric antimicrobial therapy 2012-2013.

19th ed. Elk Grove Village, IL: American Academy of Pediatrics; 2012.

2. AAP Committee on Infectious Diseases. Red Book: 2012 Report of the

Committee of Infectious Diseases. 33rd ed. Elk Grove Village, IL: Amer-

ican Academy of Pediatrics; 2012.

3. Kimberlin D. Acyclovir dosing in the neonatal period and beyond.

J Pediatr Infect Dis Soc 2013;2:179-82.

4. Dos Santos Mde F, Dos Santos OF, Boim MA, Razvickas CV, de

Moura LA, Ajzen H, et al. Nephrotoxicity of acyclovir and ganciclovir

in rats: evaluation of glomerular hemodynamics. J Am Soc Nephrol

1997;8:361-7.

5. Becker BN, Schulman G. Nephrotoxicity of antiviral therapies. Curr

Opin Nephrol Hypertens 1996;5:375-9.

6. Becker BN, Fall P, Hall C, Milam D, Leonard J, Glick A, et al. Rapidly

progressive acute renal failure due to acyclovir: case report and review

of the literature. Am J Kidney Dis 1993;22:611-5.

7. Berns JS, Cohen RM, Stumacher RJ, Rudnick MR. Renal aspects of ther-

apy for human immunodeficiency virus and associated opportunistic

infections. J Am Soc Nephrol 1991;1:1061-80.

8. Hogg RJ, Furth S, Lemley KV, Portman R, Schwartz GJ, Coresh J, et al.

National Kidney Foundation’s Kidney Disease Outcomes Quality Initia-

tive clinical practice guidelines for chronic kidney disease in children and

1468

adolescents: evaluation, classification, and stratification. Pediatrics 2003;

111:1416-21.

9. Akcan-Arikan A, Zappitelli M, Loftis LL, Washburn KK, Jefferson LS,

Goldstein SL. Modified RIFLE criteria in critically ill children with acute

kidney injury. Kidney Int 2007;71:1028-35.

10. Allison PD. Logistic Regression Using the SAS system: Theory and appli-

cation. Cary (NC): Wiley-SAS; 1999.

11. Hirji KF, Mehta CR, Patel NR. Exact inference for matched case-control

studies. Biometrics 1988;44:803-14.

12. Seedat A, Winnett G. Acyclovir-induced acute renal failure and

the importance of an expanding waistline. BMJ Case Rep

2012;2012.

13. Brigden D, Whiteman P. The mechanism of action, pharmacokinetics

and toxicity of acyclovir—a review. J Infect 1983;6:3-9.

14. Hernandez E, Praga M, Moreno F, Montoyo C. Acute renal failure

induced by oral acyclovir. Clin Nephrol 1991;36:155-6.

15. Vomiero G, Carpenter B, Robb I, Filler G. Combination of ceftriaxone

and acyclovir—an underestimated nephrotoxic potential? Pediatr Neph-

rol 2002;17:633-7.

16. Chutipongtanate S, Thongboonkerd V. Ceftriaxone crystallization and

its potential role in kidney stone formation. Biochem Biophys Res Com-

mun 2011;406:396-402.

17. Li N, Zhou X, Yuan J, Chen G, Jiang H, ZhangW. Ceftriaxone and acute

renal failure in children. Pediatrics 2014;133:e917-22.

18. Plakogiannis R, Nogid A. Acute interstitial nephritis associated with

coadministration of vancomycin and ceftriaxone: case series and review

of the literature. Pharmacotherapy 2007;27:1456-61.

19. Akgun H, Gonlusen G, Cartwright J Jr, Suki WN, Truong LD. Are

gadolinium-based contrast media nephrotoxic? A renal biopsy study.

Arch Pathol Lab Med 2006;130:1354-7.

20. Taber SS, Mueller BA. Drug-associated renal dysfunction. Crit Care Clin

2006;22:357-74. viii.

21. Sawyer MH, Webb DE, Balow JE, Straus SE. Acyclovir-induced

renal failure. Clinical course and histology. Am J Med 1988;84:

1067-71.

22. Johnson GL, Limon L, Trikha G, Wall H. Acute renal failure and neuro-

toxicity following oral acyclovir. Ann Pharmacother 1994;28:460-3.

23. Krieble BF, Rudy DW, Glick MR, Clayman MD. Case report: acyclovir

neurotoxicity and nephrotoxicity—the role for hemodialysis. Am

J Med Sci 1993;305:36-9.

24. Yavuz BB, Cankurtaran M, Halil M, Dagli N, Kirkpantur A. Renal

dysfunction after oral acyclovir treatment in a geriatric woman: a case

report. Scand J Infect Dis 2005;37:611-3.

25. Sodhi PK, Ratan SK. A case of chronic renal dysfunction

following treatment with oral acyclovir. Scand J Infect Dis 2003;

35:770-2.

26. Schreiber R, Wolpin J, Koren G. Determinants of aciclovir-induced

nephrotoxicity in children. Paediatr Drugs 2008;10:135-9.

Rao et al

Figure 1. Receiver operator characteristic (ROC) curves for A, age (weeks); B, weight (kg); and C, BMI (kg/m2) for the failuregroup. Diagonal segments are produced by ties. Values that optimized sensitivity and specificity are listed as cut points andthese were used for subsequent dichotomous comparisons in matched analyses. AUC, area under the curve.

Figure 2. Age distribution (weeks) of A, cases and B, controls.

June 2015 ORIGINAL ARTICLES

Intravenous Acyclovir and Renal Dysfunction in Children: A Matched Case Control Study 1468.e1

Figure 3. Indications for cases and controls treated with IV acyclovir based on initial presentation (admission InternationalClassification of Diseases, 9th Revision diagnoses). Proven HSV and VZV disease was based on results of polymerase chainreaction, serology, or culture results. Y-axis represents the percentage of IV acyclovir courses. GI, gastrointestinal.

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 166, No. 6

1468.e2 Rao et al

Figure 5. Percent change of eGFR from baseline plotted over time (days) for the failure group. X-axis represents days (day 1 isfirst eGFR used as baseline obtained prior to acyclovir administration). n = 14.

Figure 4. Age distribution (weeks) of cases and controls. (A-C, risk, injury, and failure cases and D-F, risk, injury, and failurecontrols).

June 2015 ORIGINAL ARTICLES

Intravenous Acyclovir and Renal Dysfunction in Children: A Matched Case Control Study 1468.e3

Table I. Cefotaxime, ceftriaxone, and vancomycin useamong cases and controls

Ceftriaxone,n (%)

Cefotaxime,n (%)

Vancomycinn, (%)

Cases,n = 131

37 (28) 44 (34) 115 (88)

Controls,n = 373

55 (15) 150 (40) 339 (91)

Table III. Rates of renal risk, injury, and failure by age

Ageclassification

n (cumulativeincidence risk

per 1000persons)

n (cumulativeincidence

injury per 1000persons)

n (cumulativeincidencefailure per

1000 persons)

<2 y 55 (162.7) 23 (68.0) 3 (8.9)2-12 y 17 (178.9) 9 (94.7) 2 (21.1)>12 y 9 (121.6) 4 (54.1) 9 (121.6)Total 81 36 14

Figure 6. Percentage of cases and controls by ceftriaxone exposure, case/control status, and age category for the renal failureanalysis. A, age #8 years; B, age >8 years. The numerals in or on the bars represent the sample size (N). Among cases andcontrols >8 years of age, ceftriaxone use was more frequent among cases than controls.

THE JOURNAL OF PEDIATRICS � www.jpeds.com Vol. 166, No. 6

1468.e4 Rao et al