This study highlights several risk factors predictive of rapid linear progression, which are uniquely expressed in different renal diseases. We also highlight distinct clinical factors associated with mortality prior to ESRD in rapid progressors compared with stable patients. Interventions targeting modifiable factors should be prioritised, especially in rapid progressors, given the significant burden of adverse outcomes experienced by this patient cohort.
Studies have shown that younger age [14], dyslipidaemia [15], lower albumin [16], lower Hb [17] and proteinuria [18] are associated with CKD progression and all these factors were predictive of patients having rapid linear progression in our analysis. Of note, we also found female gender to have a positive association with rapid linear progression. Studies that explored gender differences in CKD have found conflicting results: some found male sex confers more risk [19, 20] whereas other studies suggest the opposite [21, 22]. The exact reason for why sex differences exist in patients with CKD is not clearly understood and remains an area for further research.
We also interestingly found that higher DBP was more important than SBP in predicting rapid progression. Although historic studies have highlighted a role of DBP in progression, more recent ones have focussed on the importance of SBP alone [14], or of both SBP and DBP [23], with respect to renal outcomes. We did find higher SBP was associated with rapid progression in the univariate analysis (Additional file 1: Table S1), but it was not significant after adjustment of other covariates. This finding may be specific to our cohort but nonetheless sheds light on the need to better understand the clinical implications of DBP in those with advanced CKD, an issue recently identified by the renal community warranting further review [24].
CKD aetiology is important in predicting future progression and our study highlights the well-known association of ADPKD being most commonly linked with rapid linear progression [25]. What is perhaps less well understood is the complex interplay of factors and processes in the pathogenesis of rapid progression in other primary renal disease states. This is shown in the differential impact of exposures on three renal conditions (Table 3). For instance, rapid progressors with diabetic nephropathy were more likely to be anaemic and have A3 proteinuria, whereas rapidly progressing patients diagnosed with glomerulonephritis were more likely to have lower albumin and severe proteinuria, which is indicative of active disease and perhaps inflammation driving renal decline. Higher BMI was also associated with rapid progression in those with glomerulonephritis, but this is likely confounded by patients who were taking immunosuppressive agents such as steroids which can raise BMI.
With respect to factors associated with mortality in rapid and stable patients, there was an unsurprising representation of cardiovascular risk factors such as older age, male gender, smoking, PVD, HF and A3 proteinuria. However, these factors impacted the two patient groups in different ways. For instance, rapid progressors who had suffered a prior MI were less likely to survive, whereas there was a significant risk of mortality amongst stable patients who had suffered PVD or HF. Whether these differences are directly attributable to pathophysiological processes underlying different rates of progression requires further exploration. A3 proteinuria did not impact mortality in rapid progressors but was important for those who had stable disease. This may due to the potentially greater role severe proteinuria plays on the competing risk of ESRD in rapid progressors. Notably, use of ACEi/ARB was found to reduce the mortality risk in rapid progressors specifically. Although the beneficial effect of ACEi/ARB on mortality at different CKD stages has been highlighted in prior studies [26, 27], we show this benefit extends to those with a defined rate of rapid CKD progression. Potential protective mechanisms include favourable haemodynamic changes [28] on the cardiovascular system but also anti-inflammatory effects of renin-angiotensin-aldosterone blockade [29], which may be of particular relevance in the inflammatory milieu of rapid CKD progression.
There are several clinical implications of our findings. Firstly, there is a pressing need for accurate risk stratification that aids prognostication of adverse clinical outcomes in patients with CKD. This would not only enable targeted treatment for high-risk patients but guide appropriate interventions such as transplant referral or vascular access creation in a timely manner for those with advanced CKD [7]. Developing risk prediction calculators that take account of CKD aetiology or the rate of prior eGFR change, both of which are important determinants that influence future eGFR trajectory [30], would be desirable. Such calculators could be incorporated into electronic patient records in order to provide immediately accessible prognostic information to support clinical decision-making during outpatient consultations.
Secondly, our data clearly demonstrate that those with rapid linear progression are an especially vulnerable group of patients that suffer significantly higher annual rates of ESRD or mortality compared to their stable counterparts. Translating this to clinical practice requires assessment of patients’ rate of eGFR decline based on prior blood tests and those progressing rapidly should be offered prompt, vigorous management of modifiable risk factors and closer follow-up monitoring to mitigate future harm.
Finally, we highlight that stable CKD is also not benign. In our cohort, stable patients were older with a higher burden of cardiovascular disease, and although only 5% of patients reached ESRD, 40% of patients died. It underscores previous work showing that older patients are more likely to have stable disease, but that the absolute risk of death in this CKD subgroup remains high, largely as a consequence of cardiovascular disease [31], and this was also borne out in our study. Therefore, an equally important aspect of optimal CKD care, regardless of the rate of progression, requires addressing modifiable cardiovascular risk factors given their association with mortality [18].
There are a number of strengths to our study. Firstly, each patient had a large number of eGFR measurements taken over a long follow-up period and this helped to precisely characterise patients’ eGFR trajectories. This consequently permitted a robust analysis of patients with different rates of progression, based on their ΔeGFR slope, which was corroborated by visually inspecting each patients’ eGFR-time graphs and confirmed quantitatively by assessing the spread of the 95% CIs of the ΔeGFR in each patient group. Our systematic approach therefore ensured only patients with true CKD progression were selected. Our findings also largely support the established literature in describing key determinants of CKD progression and mortality, and in doing so also provides evidence that the phenotypic profile of those with true, linear progression is also shared with those with other rates of variable, non-linear progression described in the wider literature.
Our work also has limitations. The analysis was limited to specific ΔeGFR changes to define rapid and stable disease but did not consider the outcomes of other rates of progression, such as those between − 0.5 to -4 ml/min/1.73 m2/yr or those with larger, positive changes in eGFR over time. This latter group has also been shown to be associated with poor outcomes, perhaps related to changes in muscle mass in patients with chronic illness; or it may represent those whose trajectory is recovering from an episode of acute kidney injury, which is itself has been shown to be an independent risk factor for CKD progression [32]. Secondly, our work will be affected by limitations attributed to retrospective observational studies including an inability to confirm causal association or to account for unmeasured confounders. Thirdly, it is a single-centre study with a largely Caucasian population and thus the results may not be generalisable to other ethnic patient cohorts in other geographical locations.