In this longitudinal study, we described circadian BP patterns in RTX recipients, compared OBPM and ABPM value in this setting and identified systolic dipping status as a major determinant of kidney function.
In our cohort, prevalence of HT when combining OBPM and ABPM measurement was 82.6%. Prevalence of HT based on ABPM only was 71.9%. Those numbers are in agreement with previous studies where HT has been reported to affect as many as 80% of RTX recipients depending on BP measurement modality[22]. Non-dipping status was present in 65.8% of our patients. Again, similar studies found comparable results[12, 14]. Systolic, diastolic and mean dipping statuses were well-correlated and systolic dippers tended to be mean and diastolic dippers as well. Finally, dipping status was not associated with the presence of HT on ABPM in our cohort and dippers were as likely as non-dippers to have HT. This absence of a direct relationship between HT and dipping status was also highlighted in CKD patients where non-dippers with controlled BP were almost as prevalent as non-dippers with HT[8].
Relationship between dipping status and kidney function
Circadian BP patterns have been extensively studied in non-RTX patients. In 48 HT CKD patients followed by Timio et al., non-dippers had faster rates of renal function decline and higher proteinuria compared to dippers over a three-year follow-up[23]. In 322 patients referred for ABPM, Davidson et al. found that eGFR remained stable among dippers over a 3.2 year follow-up but declined among non-dippers independently of SBP load[4]. The same year, Agarwal et al. reported on 217 CKD patients followed during 3.5 years where non-dipping status was an independent predictor of end-stage renal disease[24]. Finally, McMullan et al. found that nocturnal dipping was associated with a decreased risk of incident CKD over a 8.1 year follow-up in 603 Afro-American patients with normal renal function[25]. As these studies globally concluded that dipping status was an independent determinant of renal function over time, some evidences suggest that such a relationship does not exist. As such, Gabbai et al. noted that, although 24 h SBP was associated with subsequent renal and CV outcomes, dipping status did not per se predict progression of renal disease in 617 Afro-American with HT CKD[26]. Likewise, two similar successive studies were published by another group enrolling 676 and 1’107 CKD patients with two and 4.7-year follow-up periods respectively[6, 8]. Authors generally concluded that non-dipping pattern with normotension did not predict the future incidence of renal outcomes.
In RTX patients, description of BP patterns is much more sparse. Successful RTX is generally thought to improve circadian BP profile in the long term[27, 28]. The impact of these profiles and renal function is however less well defined. In an early study, Kooman et al. described a relationship between nightly decrease in MAP and kidney function in 36 renal recipients[15]. Cross-sectional design and absence of adjustment for potential confounders however hampered conclusions. Haydar et al. conducted a similar study on 177 patients where SBP circadian variation was associated with eGFR[16]. The same limitations were found although a limited set of covariates was considered. Later, Wadei et al. described an association between nocturnal fall in SBP and eGFR at one year after RTX in 119 patients[17]. This study was however cross-sectional in nature and very few potential confounders were considered. The same group focused on a sub-group of 36 of these patients who had a three to four-year follow-up[29]. They found that the importance of nocturnal fall in SBP at one year was related to eGFR at last follow-up while adjusting for donor age and office SBP.
The main finding of our study is the strong and independent association between preserved systolic dipping status and improved renal function in RTX patients over a two-year follow-up. Compared to previous publications on RTX patients, our study differs on several aspects. First, owing to the longitudinal design, we implemented multi-level mixed effect analysis in order to account for repeated collection of data. By adding subject-specific random intercept and slope effects to the population average, these models permit quantification of subject heterogeneity[30]. Moreover, this methodology allows comparison of eGFR slopes, recently recognized as a valid surrogate end point and likely more useful than time-to-event analysis for short follow-up with high baseline eGFR[31]. Second, an extensive set of potential confounders was a priori selected for multivariate analysis. This is of prime importance in this field as several factors, particularly BP control and proteinuria, were shown to confound the intricate relationship between dipping status and renal function. As an example, Agarwal et al. found that non-dipping status was predictive of CV events in a prospective cohort of 217 CKD patients[7]. However, this relationship disappeared when adjusted for proteinuria or clinic BP. In our study, a preserved dipping status was associated with a higher eGFR independently of potential confounders. The presence of HT was the only other significant determinant of renal function in longitudinal analysis. However, it is worth noting that not only did dipping status maintain its association with renal function beyond intensity of BP control but also its effect size on eGFR was comparable to that of HT itself. A recent study by Mallamaci et al. was designed similarly to ours as the authors reported on the impact of baseline ABPM on subsequent renal function over a 3.7-year follow-up in a cohort of 260 RTX recipients[18]. In multivariate analysis, 24 h, daytime and night-time absolute BP values were negatively associated with eGFR. However, this crude description of main ABPM components gave no specific information about a potential impact of circadian BP patterns or dipping status. Moreover, although presented analyses account for elapsed time, eGFR slopes are not depicted. The purpose of this study was thus rather different than ours and results are not directly comparable.
Finally, in previous studies, whether alteration in circadian BP profile was the cause or the consequence of kidney function decline was not entirely clear. In our study, although dippers had higher eGFR compared to non-dippers overall, differences seemed most striking at T1 when ABPM was recorded. eGFR slope analysis confirmed that dippers had a faster rate of function decline compared to non-dippers in univariate analysis. When considering potential confounders however, kidney function decline rates were similar between groups. This type of time trajectory is in agreement with the underlying physiological assumption that correlation between dipping status on a single ABPM assessment and induced TOD should be maximal at the time of initial measurement to then decrease during follow-up owing to the low reproducibility of circadian patterns over time[32]. As a matter of fact, any inter-group comparison in a follow-up study is based on dipping classification defined at baseline ABPM that does not necessarily represent actual dipping status at later time points. In a previous study on CKD patients with established diabetic nephropathy, time trajectories of renal function according to dipping status were similar to ours[33]. Globally, those findings are thus in favour of a causal effect of dipping status on renal function and not the other way around. It has to be noted that Wadei et al. described the opposite phenomenon in their longitudinal study where the association between dipping status and renal function was inexistent at initial evaluation but became significant during follow-up[29]. This apparent paradox is however not incompatible with our results. First, compared to our study, minimal adjustment only was considered and those results could represent residual confounding. Second, ABPM was conducted one-year post transplant in their cohort while median graft vintage at ABPM was 2.5 years in our study. Thus, the longer time span between transplant and initial evaluation in our study could have allowed sufficient influence of established circadian BP patterns to impact renal function.
Limitations
Our study has limitations that should be considered when interpreting the results. First, the observational nature of the design hampers definitive conclusions on causal relationship between considered variables. Namely, as the interplay between dipping status (and more generally BP control) and renal function is intricate, reverse causality should be considered. We believe however that the longitudinal design, the statistical methodology as well as the pathophysiologic considerations help comfort our conclusions. Second, as in most similar studies, a single ABPM measurement was conducted. Longitudinal evaluation could thus be hampered by the intrinsic moderate reproducibility of dipping categorization. We believe however that our findings were discussed in the light of this phenomenon. Third, the rather limited sample size prevented us to further investigate specific sub-group of patients and to detect potential association with smaller effect size. Statistical power was however sufficient to investigate our primary objective as attested by the highly significant and robust associations presented. Fourth, as a fraction only of screened RTX patients had ABPM performed, selection bias cannot be excluded. Excluded patients were however similar to those included. Finally, the number of donor characteristics considered in our study was limited, as ethics regulation does not allow documentation of such data in medical records at our institution.