We identified 30 eligible patients with ARPKD and available echocardiogram data from 2000 to 2022, 11 of whom had ABPM data. Genetic testing results were available in 20 patients, all of whom had at least one PKHD1 variant detected. Of these, five patients had homozygous PKHD1 variants, and 13 patients had compound heterozygous PKHD1 variants. Three patients who had not undergone genetic testing had a full sibling with a genetically confirmed ARPKD diagnosis. Thus, 23 of 30 patients (77%) had genetic evidence of ARPKD. In the full cohort, the median age at the time of echocardiogram was 7.2 (range 0.2–19.4) years, with a median eGFR of 49 mL/min/1.73 m2. In the ABPM cohort, median age at the time of echocardiogram was 11.5 years, with a median eGFR of 45 mL/min/1.73m2 (Table 1).

In the full cohort, median casual SBP was at the 77th percentile, median casual DBP was at the 71st percentile, and median hypertension score was 1 (Table 1). Casual SBP was positively correlated with casual DBP (r = 0.80, P < 0.001; Supplementary Table 1). In the ABPM cohort, median casual SBP was at the 64th percentile and median casual DBP was at the 40th percentile; median SBP and DBP indexes during wake and sleep ranged from 0.90 to 0.93, as shown in Table 1. ABPM SBP indexes during wake and sleep were positively correlated (r = 0.70, P = 0.02); similarly, DBP indexes during wake and sleep were positively correlated (r = 0.71, P = 0.02; Supplementary Table 1). ABPM SBP index was not significantly correlated with DBP index during wake or sleep (Supplementary Table 1). ABPM wake DBP index was positively correlated with casual DBP percentile (r = 0.71, P = 0.02) but there were no other significant correlations between ABPM SBP and DBP indexes and casual BPs. Hypertension score was positively correlated with ABPM sleep DBP index (r = 0.76, P = 0.006) but was not significantly correlated with any other ABPM indexes (Supplementary Table 1).

Casual SBP and DBP percentiles were both significantly negatively correlated with age (r = − 0.74, P < 0.001, and r = − 0.81, P < 0.001, respectively; Supplementary Table 1 and Fig. 1A, B). ABPM wake DBP index was negatively correlated with age (r = − 0.61, P = 0.048), but remaining ABPM indexes were not significantly correlated with age (Supplementary Table 1). Neither casual BP percentiles nor ABPM indexes were significantly correlated with eGFR (Supplementary Table 1).

Relationship between age, casual blood pressure (BP), hypertension score, and left ventricular geometry in patients with autosomal recessive polycystic kidney disease (ARPKD). A Scatterplot of age at echocardiogram and casual systolic BP (SBP) percentile; B Scatterplot of age at echocardiogram and casual diastolic BP (DBP) percentile; C Scatterplot of age at echocardiogram and hypertension score; D Scatterplot of age at echocardiogram and left ventricular (LV) mass Z-score; E Box plot of age at echocardiogram stratified by presence of left ventricular hypertrophy (Khoury et al. criteria, [20]); F. Box plot of age at echocardiogram stratified by presence of left ventricular hypertrophy (Chinali et al. criteria, [19])

Most patients [93% (26/28; missing medication data in n = 2)] were on antihypertensive medications. Combinations of classes of antihypertensive therapy used in patients at the visit closest to the echocardiogram are shown in Supplementary Fig. 1. ACE-I was the most common class of medications used (n = 17 patients total, as monotherapy in n = 12 patients). The number of antihypertensive medications was negatively correlated with age at echocardiogram (ρ = − 0.46, P = 0.014) and positively correlated with eGFR (ρ = 0.43, P = 0.022) but was not significantly correlated with casual SBP or DBP percentiles (ρ = 0.28, P = 0.15 and ρ = 0.30, P = 0.11, respectively) or ABPM SBP or DBP indexes (ranges of ρ = −0.31 to 0.41, P = 0.21 to 0.49). Five of the 26 patients receiving antihypertensive therapy (19%) had uncontrolled hypertension (i.e. SBP or DBP > 95th percentile (Table 1)). Hypertension score was negatively correlated with age (ρ = −0.53, P = 0.004, Fig. 1C); there was a trend towards negative correlation with eGFR, but this did not reach statistical significance (ρ = −0.37, P = 0.06; Supplementary Table 1).

In the full cohort, 23% of patients (7 of 30) had LVH by Khoury et al. criteria [20], with median LVMI of 40.5 g/m2.7. Using Chinali et al. criteria [19], 50% of patients (15 of 30) had LVH, with median LVMI of 47.2 g/(m2.16 + 0.09). Median LV mass Z-score was 0.54 (Table 1).

LVMI in g/m2.7 was negatively correlated with age (r = −0.48, P = 0.01), while LVMI in g/(m2.16 + 0.09) was not significantly correlated with age (r = −0.033, P = 0.87), consistent with observations in healthy reference populations [19, 20](Supplementary Table 1). LV mass Z-score was not significantly correlated with age (r = −0.04, P = 0.83, Fig. 1D). Children who had LVH by Khoury et al. criteria [20] were significantly younger than those without LVH (median 1.76 vs. 9.08 years, P = 0.02, Fig. 1E), but there was no significant group difference in median age by LVH status when LVH was classified by Chinali et al. criteria [19](Fig. 1F). LV mass Z-score was significantly higher in males than in females (1.03 ± 1.63 vs. −0.54 ± 1.06, P = 0.005).

Comparisons of patients with and without LVH by Khoury et al. criteria [20] showed no significant differences in overall mean casual SBP and DBP percentiles (SBP percentile 73.5 vs. 77.5, P = 0.9; DBP percentile 87.4 vs. 67.5, P = 0.2, Fig. 2A). ABPM parameters were also not significantly different between patients with and without LVH (wake SBP index 0.93 vs. 0.93, P = 0.9; wake DBP index 0.87 vs. 0.90, P = 0.9; sleep SBP index 0.94 vs. 0.92, P > 0.9; sleep DBP index 0.90 vs. 0.92, P = 0.7; Fig. 2B, C). Comparisons of patients with and without LVH by Chinali et al. criteria [19] similarly showed no significant differences in overall mean casual BP percentiles or ABPM indexes (Supplementary Fig. 2).

Relationship between left ventricular hypertrophy (LVH) by Khoury et al. criteria [20], left ventricular (LV) mass Z-score, and blood pressure (BP) in patients with autosomal recessive polycystic kidney disease (ARPKD). A Box plot of casual systolic BP (SBP) and diastolic BP (DBP) percentiles in patients with and without LVH; B Box plot of ambulatory blood pressure monitor (ABPM) wake SBP and DBP indexes in patients with and without LVH; C Box plot of ABPM sleep SBP and DBP indexes in patients with and without LVH; D Scatterplot of casual SBP percentile and LV mass Z-score; E Scatterplot of casual DBP percentile and LV mass Z-score. F Scatterplot of hypertension score and LV mass Z-score

There were no significant linear correlations between LV mass Z-score and casual SBP or DBP percentiles (r = −0.08, P = 0.7; and r = −0.35, P = 0.07, respectively; Figs. 2D, E). The number of antihypertensive medications was positively correlated with LVMI in g/m2.7 (ρ = 0.55, P = 0.003), but was not significantly correlated with LVMI in g/(m2.16 + 0.09) (ρ = 0.32, P = 0.10) or LV mass Z-score (ρ = −0.29, P = 0.2). Hypertension score was positively correlated with LVMI in g/m2.7 (ρ = 0.52, P = 0.005) but was not significantly correlated with LVMI in g/(m2.16 + 0.09) (ρ = 0.16, P = 0.4) or LV mass Z-score (ρ = 0.18, P = 0.4) (Supplementary Table 1).

In univariate linear regression analysis, LVMI in g/m2.7 was positively associated with hypertension score and LVMI in g/(m2.16 + 0.09) was negatively associated with mean casual DBP percentile. There were otherwise no significant associations between LVMI [in either g/m2.7 or g/(m2.16 + 0.09)] or LV mass Z-score and any other BP predictor variables, including casual SBP and DBP percentiles and ABPM BP indexes (Supplementary Table 2).

After adjustment for age, sex, and eGFR, linear regression analysis showed that LV mass Z-score was negatively associated with casual DBP percentile (β coefficient −0.31, P = 0.04) but there were otherwise no significant associations between LV geometry outcomes and any BP predictor variables (Supplementary Table 2).

There were no significant linear correlations between eGFR and age, casual BPs, ABPM BP indexes (Supplementary Table 1).

Patients with LVH by Khoury et al. criteria [20] had a significantly lower eGFR than those without LVH (21 vs. 65 mL/min/1.73 m2, P = 0.02, Fig. 3A) but there was no significant difference in eGFR in patients with versus without LVH by Chinali et al. criteria [20] (45 vs. 65 mL/min/1.73m2, P = 0.18, Fig. 3B). Significant negative correlations were found between eGFR and LVMI in g/(m2.16 + 0.09) (r = −0.43, P = 0.02, Supplementary Table 1) and between eGFR and LV mass Z-score (r = −0.41, P = 0.03, Fig. 3C). In unadjusted univariate linear regression, both LVMI in g/(m2.16 + 0.09) and LV mass Z-score were negatively associated with eGFR (β coefficient −0.84, P = 0.02 and β coefficient −0.07, P = 0.03, respectively; Supplementary Table 2). After adjusting for age and sex, both LVMI [in g/m2.7 and g/(m2.16 + 0.09)] and LV mass Z-score were found to be negatively associated with eGFR (β coefficient −1.45, P = 0.03; β coefficient −0.82, P = 0.01; and β coefficient −0.08, P = 0.02, respectively; Supplementary Table 2). After further adjustment for mean casual SBP and DBP percentiles, in addition to age and sex, there were still significant negative associations between both LVMI [in g/m2.7 and g/(m2.16 + 0.09)] and LV mass Z-score and eGFR (β coefficient –1.08, P = 0.003; β coefficient –0.79, P = 0.007; and β coefficient −0.07, P = 0.01, respectively; Supplementary Table 2). Similar negative associations were found between both LVMI and LV mass Z-score and eGFR after adjustment for age, sex, and hypertension score (Supplementary Table 2).

Relationship between left ventricular geometry and estimated glomerular filtration rate (eGFR). A. Box plot of eGFR stratified by left ventricular hypertrophy (LVH) status (Khoury et al. criteria, [20]); B Box plot of eGFR stratified by presence of LVH (Chinali et al. criteria, [19]); C Scatterplot of eGFR by LV mass Z-score