Autosomal recessive polycystic kidney disease subjects and controls were selected from among those enrolled in CKiD, a longitudinal, prospective study of children with mild-to-moderate CKD. Complete eligibility criteria for CKiD have been published in detail elsewhere (7) and include age 1–16 years, estimated GFR of 30–90 mL/min/1.73 m² calculated using the original Schwartz formula (8, 9) at enrollment, and absence of severe syndromic disease. The CKiD study protocol was approved by the Institutional Review Boards at all participating sites, and informed consent was obtained from all caregivers.

The current study was performed as a control-matched analysis within the CKiD cohort. All children enrolled in CKiD with a diagnosis of ARPKD were included in this analysis. Controls were obtained from two diagnostic groups with other congenital renal diseases: (1) aplastic/hypoplastic/dysplastic (A/H/D) kidneys and (2) obstructive uropathy (OU). Controls were matched to ARPKD subjects in a 2:1 ratio based on baseline ieGFR [GFR measured by plasma disappearance of iohexol (iGFR) or eGFR (10), if iGFR not available], age at study entry, and age at diagnosis. Matching was performed by calculating the Euclidean distance in 3-dimensional space (with baseline ieGFR, age at study entry, and age at diagnosis being the dimensions) between each of the 22 ARPKD subjects and each potential control (all subjects in CKiD with A/H/D or OU diagnoses, with each control group matched separately). The smallest distance (closest match) was then selected out of all possible matches; that control was removed, and then the procedure was repeated to choose closest matches until each ARPKD subject had two distinct matches.

Height or length was determined at all CKiD study visits as the mean of two independent measurements, using a stadiometer for children >2 years old who are able to stand, or using a firm box in the supine position for children unable to stand (7). Age- and sex-specific height SD scores (z-scores) were calculated using United States (US) Centers for Disease Control and Prevention (CDC) growth charts of normative data (11). Growth hormone usage was ascertained at each visit using a standardized medication questionnaire. GFR was estimated yearly using the updated biomarker-based Schwartz formula (eGFR) (10) and was measured every other year using plasma disappearance of iohexol (iGFR) (12).

The primary outcomes examined in this study were height z-score, proportion of children with severe short stature (height z-score < −1.88), rates of growth hormone use at baseline, and annual change in height z-score. We also examined other covariates that could have an influence on linear growth, including weight and body mass index (BMI, as indicators of nutritional status), pubertal (Tanner) stage, serum bicarbonate concentrations, and use of bicarbonate supplementation [to evaluate degree of metabolic acidosis, which previous studies have shown can affect linear growth (13)], intact parathyroid hormone (iPTH) concentrations and use of phosphate binders (to assess for secondary hyperparathyroidism), hemoglobin [some studies have identified anemia as a risk factor for poor growth in CKD (14)], platelet count [to assess severity of portal hypertension in children with ARPKD (15)], albumin (to assess liver synthetic function), and need for intensive care unit and days of hospitalization at birth (to assess for severity of other comorbidities such as pulmonary hypoplasia). Additional data regarding neonatal course, other comorbidities, growth hormone levels, or IGF binding protein levels were not available in this cohort. As an exploratory analysis, we also compared annual change in height z-score between users and non-users of growth hormone in all three groups.

Demographic and clinical characteristics were reported as median (interquartile range, IQR) or number, n (%), for each group. Differences between the ARPKD group and each control group were tested by Wilcoxon rank-sum test or Fisher’s exact test. Annualized change in height z-score was calculated using individual regressions for each subject incorporating all available follow-up measurements. Matched differences were calculated as the difference between an ARPKD subject’s value and the average value of the two matched controls, and these distributions were tested for difference from 0 by Wilcoxon signed-rank test.

Since the sample size of ARPKD patients enrolled in CKiD was fixed, we did not perform power calculations to determine sample size. However, given the available sample size (n = 66 total, with n = 22 ARPKD and n = 44 total controls), we would have been able to detect a difference in baseline height z-score of 0.83 or a difference in annual change in height z-score slope of 0.16.

The baseline demographic and clinical features of the 22 ARPKD subjects and the A/H/D and OU control groups (n = 44 each) have been reported in our previous publication examining kidney disease progression in ARPKD subjects in the CKiD cohort (16). The ARPKD and control groups (A/H/D and OU) were successfully matched on baseline ieGFR, age at study entry, and age at diagnosis. For the ARPKD cohort, the mean age at study entry was 7.9 years and the mean baseline ieGFR was 43 mL/min/1.73 m². Twenty-seven percent of ARPKD patients were premature (<36 weeks gestation), and 18% had low birth weight (<2500 g). There were no significant differences between the ARPKD subjects and the two matched control groups for any of these parameters.

Baseline growth characteristics of the ARPKD group and the A/H/D and OU control groups are shown in Table 1. All three groups had height z-scores below the normal range compared with normative data for US children, but there were no significant differences in median height z-score among the three groups (ARPKD −1.1 vs. A/H/D −0.8, p = 0.97, ARPKD −1.1 vs. OU −0.6, p = 0.63). A large proportion of children in all three groups had severe short stature (height z-score less than −1.88, i.e., <3rd percentile), but there were no significant differences between the three groups (ARPKD 14% vs. A/H/D 19%, p = 0.74, ARPKD 14% vs. OU 23% p = 0.52). The proportions of children using growth hormone were also similar between the three groups (ARPKD 18% vs. A/H/D 9%, p = 0.42, ARPKD 18% vs. OU 14%, p = 0.72).

Examination of other covariates showed no significant differences between the ARPKD group and the A/H/D and OU control groups in weight and BMI z-scores, pubertal (Tanner) stage, serum bicarbonate concentrations, use of bicarbonate supplementation, iPTH concentrations, and use of phosphate binders. The ARPKD group had lower median hemoglobin than the A/H/D and OU control groups (ARPKD 11.7 g/dL vs. A/H/D 12.6 g/dL, p = 0.001, ARPKD 11.7 g/dL vs. OU 12.6 g/dL, p = 0.001) (Table 1).

We assessed for severity of portal hypertension in the ARPKD group by comparing platelet counts across the groups. Although there were no significant differences in median platelet counts between the ARPKD and A/H/D and OU control groups (Table 1), a subset of the ARPKD group (6 out of 22, 27%) did have clinical evidence of portal hypertension, as evidenced by thrombocytopenia (platelet count < 150,000/μL). Median height z-score at baseline in ARPKD patients with thrombocytopenia (n = 6) was −0.39 [IQR −1.47, −0.16], compared with −1.28 [IQR −1.48, −0.27] in those with normal platelet counts (n = 16). Although the small number of subjects precludes statistical analysis, the similarity of the interquartile ranges in the two groups suggests that there was no significant effect of portal hypertension on baseline height. Serum albumin was similar between the ARPKD and control groups, indicating normal liver synthetic function (Table 1).

Although detailed clinical information regarding the presence of pulmonary hypoplasia at birth was not available, the ARPKD group had similar prevalence of needing ICU care at birth and similar durations of hospitalization compared to the A/H/D and OU control groups (Table 1).

Observed values for annual change in height z-score were −0.02 [IQR −0.16, 0.10] in the ARPKD group, −0.02 [−0.11, 0.08] in the A/H/D group, and −0.01 [−0.07, 0.08] in the OU group (Figure 1). Examination of matched differences in annual change in height z-score showed no significant differences between the ARPKD and A/H/D groups (p = 0.52) or between the ARPKD and OU groups (p = 0.54).

We also explored the response to growth hormone in the ARPKD and control groups by examining annual change in height z-score in users and non-users of growth hormone. Although the small sample size precludes statistical analysis, it is interesting to note that the ARPKD subjects who were receiving growth hormone (n = 4) at baseline had lower annual change in height z-score (median −0.038) compared with those with A/H/D (n = 4, median 0.097) or OU (n = 6, median 0.034) (Table 1).