Introduction

In chronic kidney disease (CKD), when the glomerular filtration rate (GFR) approaches the level of 15 ml/min/1.73 m2 [stage 5 or end-stage renal disease (ESRD)], nephrologists face the difficult choice between starting renal replacement therapy (RRT) or prolonging conservative therapy. RRT represents, at least in higher- or middle-income countries, the “easy and safe” choice; however, consensus is now growing on conservative management (CM) as a treatment alternative to RRT in ESRD [1].

The recent KDIGO Controversies Conference on Supportive Care in CKD identified conservative therapy as a priority for improving patient-centered care [2]. Indeed, the opinion of the Conferees on the best treatment for ESRD—whether CM or RRT—was far from unanimous, with some claiming that RRT improves survival in the large majority of patients while others stated that it may offer limited, if any, benefit in terms of survival or quality of life. Similarly, in a survey among European nephrologists on the decision-making process, while level of renal function emerged as the most important factor in uncomplicated patients, several other factors were reported as needing to be taken into account in the population at large [3]. Therefore, more studies are needed to help clarify decision-making about when CM is still appropriate in ESRD [2, 4].

We performed a systematic review of studies published in the last decade comparing the main clinical outcomes of stage 5-CKD patients under CM versus RRT. The aim of this analysis was to expand knowledge on the best treatment of ESRD, given the low number of studies published on this issue and the fact that only one systematic analysis is available in the field (though not specifically aimed at comparing CM versus RRT in ESRD and, moreover, updated only to 2011) [5]. The information derived from the systematic analysis was integrated with a case report of a patient from our outpatient clinic with a prolonged course of non-dialysis stage 5-CKD.

Systematic review

Methods

We searched published studies comparing clinical outcomes, i.e. rates of survival, hospitalization, and quality of life, in stage 5-CKD patients under conservative versus dialysis therapy. A systematic search of articles published in all languages was performed using PubMed, including Medline, Scopus and Web of Sciences databases, to identify relevant studies published in the last decade (January 2005 to September 2015). We used the following medical subject headings (MeSH) and text words: “nondialytic”, “non-dialytic”, “conservative management”, “conservative”, “conservatively” and “chronic kidney failure”, “chronic renal failure”, “end-stage renal disease”, “end-stage kidney disease”, “end-stage renal failure”, “stage 5 CKD”, and “advanced CKD”. The detailed search syntax is reported in Appendix (Item S1). Bibliographies of relevant articles and reviews were also manually screened for additional studies.

Original studies, either randomized or observational, comparing conservative versus dialysis treatment in stage 5-CKD, were retained; studies were excluded if sample size was less than 50 patients and/or no data on the outcomes of interest were provided. The search was designed and performed by one author (CG) under the supervision of two senior authors (GC and LDN).

Results

A flow diagram of the selection process is presented in Fig. 1. Eleven studies were finally considered for analysis (Table 1).

Fig. 1
figure 1

Flow diagram of the literature selection process

Full size image
Table 1 Studies in the last decade comparing outcomes of stage 5-CKD patients under conservative versus dialysis treatment
Full size table

We found only two randomized clinical trials (RCT). The Diet Or Dialysis in the Elderly (DODE) was a multicenter trial in Italy showing that a supplemented very low protein diet (0.3 g/kg body weight) can safely postpone dialysis treatment for about one year in elderly patients (aged > 70 years) with very low GFR (5–7 ml/min/1.73 m2) [6]. The main limitations were exclusion of diabetics and incomplete enrollment that reduced the statistical power of analysis. The other study, Initiating Dialysis Early and Late (IDEAL), enrolled quite a large number of adult patients (mean age 60 ± 12 years) to compare outcome of early versus late dialysis start [7]. Results showed that dialysis may be safely delayed with careful clinical management. However, 19 % of subjects of the “early start” group commenced dialysis later while 76 % of “late start” subjects started early. Hence, the difference in eGFR at the time of initiation of dialysis was smaller than originally planned (eGFR: 9.0 and 7.8 ml/min/1.73 m2 in the early and late group, respectively), and the median difference in time to dialysis initiation was only 5.6 months. Of note, the vast majority of patients had been seen by the nephrologist for about 10 months before enrollment; IDEAL patients in fact appeared well nourished and well prepared, with an extraordinarily high percentage (60 %) starting RRT by peritoneal dialysis (PD) modality.

The nine observational studies included elderly patients in the vast majority of cases, with patients under conservative treatment being of a consistently higher age compared to the dialysis group [8–16]. Besides the advanced age, these studies had major methodological drawbacks. A complete description of essential basal features, such as prevalence of diabetes and cardiovascular disease (CVD), was in fact provided in only two studies [14, 15]. Furthermore, most (n = 7) studies had a retrospective design [8–10, 13–16]. Finally, follow-up was generally short with only three studies having a median value longer than 4 years [9, 10, 15]. In interpreting the results of these studies, therefore, these methodological pitfalls should be borne in mind.

The retrospective study by Carson et al. showed that dialysis prolongs survival by approximately 2 years in elderly patients who have ESRD and significant comorbidities [9]; however, this result was obtained in a small CKD population that started hemodialysis (HD) after only short-term (<3 months) or no nephrology care in almost half of the cases, and, more important, survival analyses were not adjusted for potential confounders. Conversely, two studies with adequate sample size and a longer follow-up did not confirm the superiority of RRT over CM in fully adjusted analyses [10, 15].

The importance of adjusting for comorbidities further emerges in the study by Murtagh et al. [8]. This study showed that in patients over 75 years of age, and who received nephrology care early, the survival advantage of dialysis was significant, with a 2-year survival rate of 76 versus 47 % in dialysis and CM, respectively; however, the survival advantage was lost in patients with high comorbidity scores, particularly those with ischemic heart disease. Similarly, in the study by Hussain et al. [13], RRT showed a survival advantage over CM but, again, as age and co-morbidity burden increased and performance status deteriorated, dialysis lost its survival advantage. Of note, RRT patients were also more frequently admitted acutely to, and died in, hospital [13]. A recent study by Shih et al. confirmed that age is a potent modifier of the association between dialysis therapy and mortality risk [16]. The authors evaluated a large population (>8000 patients aged ≥70 years) with advanced CKD (eGFR < 15 ml/min/1.73 m2) treated either conservatively or dialyzed. They found that dialysis therapy was associated with a 40 % higher mortality risk compared to conservative care [adjusted hazard ratio (HR) 1.39, 95 % confidence interval (CI) 1.30–1.49] [16]; in subgroup analyses, moreover, the mortality risk in dialysis remained consistently increased, independently of sex, comorbidities and even age.

The major role of disease severity in modifying prognosis of dialysis-treated patients was found by Shum et al. when comparing CM vs. peritoneal dialysis [14]; they found that the survival advantage of dialysis was preserved in patients receiving PD with low comorbidity but was lost in those with high comorbidity. Similarly, Da Silva et al. showed that mortality risk in the dialysis group was approximately half that in CM patients [HR 0.47 (0.20–1.10); p = 0.08] but the difference was less in patients with high comorbidity score [11]. As for survival, also health-related quality of life seems not to improve in dialysis-treated ESRD patients who are elderly or have a high comorbidity burden [12].

Comments

The progressive nature of CKD has always been a key research focus in Nephrology [17, 18]. In 1991, Maschio et al. hypothesized the existence of a ‘point of no return’, corresponding to a residual GFR of about 50 ml/min beyond which progressive worsening of renal function is considered irreversible and poorly responsive to any nephroprotective intervention [19]. Recent studies have challenged this traditional view by revealing different rates of GFR decline in CKD of similar severity, with some patients experiencing a rapid progression to RRT and others maintaining stable renal function over time independently of the disease stage [20]. Indeed, non-progressive CKD is typically encountered in the subgroup of older non-diabetic patients with low proteinuria and mild to moderate disease (CKD stage 1–3) [21–26]; nonetheless, non-progressors have been identified, though less frequently, in more advanced CKD [27–30]. The recent paradigm shift on the “equivalence” between CKD and unrelentless decline of GFR is, at least in part, dependent on the wider implementation of preventive nephrology care with its multifaceted therapeutic armamentarium aimed at limiting progression to RRT and mortality [31–34].

The persistent, and critical, dilemma mainly derives from the absence of solid evidence-based studies on this issue. Previous observational single-arm studies have shown that conservative care by the nephrologist can prolong for 6–24 months the dialysis-free condition safely and with satisfactory quality of life, in patients with stage 5-CKD [1, 5, 35–39]. Patients included in these studies were mostly over 75 years of age; the only exception was the study by Di Micco et al. in 30 middle-aged patients (55.7 ± 13.4 years) [37] showing that nephrology care was associated with a median time of 11.8 months from eGFR ≤ 11.0 ml/min/1.73 m2 to dialysis. On the other hand, the validity of HD as an exclusive therapy for ESRD for all patients has been confuted by a large study (n = 3702) in nursing home residents starting treatment in the United States (aged 73.4 ± 10.9 years, basal eGFR 10.7 ± 4.9 ml/min/1.73 m2) [40]; the authors found that only 25 % survived and only 1/8 of survivors maintained functional status after the first year of dialysis.

Our systematic review, though limited by the search made by a single author, adds novel information on this issue by specifically examining, for the first time, the studies comparing outcomes of stage 5-CKD patients under RRT versus CM (Table 1). In interpreting the results of the studies, one should consider major drawbacks, including the highly selected study population with patients aged over 70 years in most cases. Overall, the studies do not indicate a superiority of dialysis, especially in the presence of older age and/or high comorbidity scores, in terms of survival or of quality of life. These findings therefore highlight the possibility of safely delaying RRT in advanced CKD by careful nephrology care in very ill patients as well as in those very old.

Case report

Our patient is a Caucasian woman born in 1947, with type 1 diabetes mellitus (DM1) diagnosed in 1974. In 1994, we transitorily followed her because of renal tuberculosis (TB)-induced obstructive acute kidney injury (AKI). Triple treatment of TB for 1 year allowed eGFR to recover to basal value within 6 months, as previously described [41]. At that time, the right kidney was surgically removed as gross cavities, in which the renal parenchyma was a thin shell, were documented. Subsequent controls to detect Mycobacterium in the urine were all negative while several bacterial urinary tract infections (UTIs) were diagnosed and treated.

In 1999, she was referred by the general practitioner (GP) to our clinic because of low eGFR (17.8 ml/min/1.73 m2). Since then, the patient has been regularly followed in our outpatient clinic (Fig. 2), with several visits and biweekly-to-monthly phone contacts. Despite her low educational level (only primary school), the patient was efficaciously instructed as regards recognizing symptoms of UTI and volume depletion, and monitoring body weight and blood pressure (BP). The main clinical and laboratory parameters during follow-up are reported in the Appendix (item S2).

Fig. 2
figure 2

MDRD-GFR and main events during nephrology follow up

Full size image

Treatment was multifactorial and aimed, from commencement of the nephrology follow-up, at optimizing metabolic and hypertension control. Overall, the therapy was maintained throughout follow-up (Table 2), while drug dosing was frequently modified (in 53 % of the 125 visits performed from 1999 to 2015). Adherence to dietary recommendations was adequate, with protein and salt intake consistently ranging 0.6–1.0 g/kg/day and 5–10 g/day, respectively, as testified by the multiple measurements of 24 h urinary excretion of urea and sodium.

Table 2 Prescribed therapy and dosage in the last 3 months (April to June 2015)
Full size table

Repeated obstructive AKI episodes due to left ureteral stenosis required permanent percutaneous nephrostomy in 2001. These episodes, and complications related to surgery, caused a 10 ml-decrement of eGFR from 17.4 in 1999 to 7.8 ml/min/1.73 m2 in 2004. In the subsequent decade, eGFR remained remarkably stable, with a total eGFR loss of about 1.0 ml/min. In 2002, the low eGFR levels led us to prepare the patient for hemodialysis as she refused PD and was ineligible for preemptive transplantation; two consecutive native arteriovenous fistulae (AVF) were therefore prepared but both of them failed to mature. On the other hand, the stability of her clinical and metabolic parameters, and the immediate prevention of acute complications such as volume expansion or hyperkalemia, allowed us to safely avoid acute RRT. The patient did not develop an impaired nutritional status and/or inflammation, as evidenced by persistent normality of body mass index (BMI), ferritin, leucocytes count, and albumin.

During the 2004–2012 period, associated with the slight decline of eGFR from 8 to 6 ml/min, proteinuria decreased from 2.5 to 0.6 g/24 h, on average. The reduction of proteinuria was, at least in part, independent of the modest eGFR decline; the two parameters in fact did not correlate (n = 42, r = −0.127, p = 0.423) and the proteinuria/eGFR ratio halved (from 0.22 ± 0.15 to 0.10 ± 0.05 g/ml eGFR). Thereafter, 24-h urine collection was no longer possible, because the patient became blind at the end of 2012 due to the irreversible worsening of diabetic retinopathy.

At the last control (May 2015), the patient was free of asthenia, dyspnea, and peripheral and pulmonary edema; BP was 144/50 mmHg without orthostatism, heart rate 84 b/m, eGFR 5.6 ml/min/1.73 m2, sK 4.2 mEq/L, HCO3 24 mEq/l, urea 190 mg/dl. In the previous month, a computed tomography (CT) of the thorax showed a calcified granuloma and multiple calcified lymphadenopathies in the left lung, diffuse interstitial thickening in both lungs without cavity or pleural effusion. Echography showed that the left kidney was small with no dilation of pelvis and correct location of nephrostomy. Vascular calcifications were evident in the main abdominal aortic branches. Echocardiography concomitant to the last visit showed left ventricular hypertrophy (77 g/h2.7) with preserved ejection fraction (70.4 %) and a normal inferior cava vein (diameter 1.3 cm, collapsibility index 69.2 %). Diastolic function showed a type II pattern (pseudonormal, E 6 cm/s and E/e′ 21.3).

A few days after the last control visit, the patient developed oliguric AKI because of persistent hyperglycemia, ascribed to repeated errors in insulin dosing, with dependent volume depletion; unfortunately, renal function did not recover after medical therapy and she started chronic RRT, at the age of 68 years, after a period of 15 years spent in CKD stage 5.

Comments

Our systematic review of the literature suggests that the nephrologist intervention allows to safely prolong, for up to 4 years on average, dialysis-free survival in older patients with advanced CKD. Our case report may extend the efficacy of conservative management to both long-term (15 years) and younger (middle-aged) patients.

The striking feature of this case report is the prolonged regular follow-up in Nephrology care (16 years, 125 visits). We therefore had the opportunity to observe the maintenance of eGFR, even if remarkably low, for an extended period of time (about 15 years at CKD-stage 5). To our knowledge, this is the longest stability of “pre-dialysis” renal function ever described. Notably, this remarkable stability was observed in a patient at high risk of ESRD because of her age (she was 52 years old when she started nephrology care with eGFR < 20 ml/min), had long-standing DM1 with microvascular damage (retinopathy leading to blindness), severity of renal disease (single kidney with nephrostomy), and multiple intercurrent AKI episodes.

Histological characterization of renal damage was not obtained because the single functioning kidney contraindicated biopsy. Furthermore, we did not measure true GFR; however, the Modification of Diet in Renal Disease (MDRD)-eGFR shows an adequate performance in advanced CKD [42]. This holds particularly true in the absence of under-nutrition [43], as in the case of this patient. The enduring severity of disease was further supported by the multiple 24-h creatinine clearance measurements obtained, which were similar to eGFR values and consistently below 15 ml/min.

It is reasonable to hypothesize that the intensive nature of the nephrology care may have played a major role in determining the non-progressive status. In this regard, we recently evidenced in a cohort of 729 CKD patients with advanced disease (79 % in stage 3–4), selected from our outpatient renal clinic on the basis of availability of diagnosis of underlying renal disease, that adequate control of hypertension, anemia and proteinuria in the first year of nephrology care halves the risk of ESRD over the subsequent 5 years [34]. Of note, this association was independent of the nature of primary renal disease, suggesting that the heterogeneous risk of progression, which is intrinsic to each type of underlying disease, may be homogeneously reduced by nephrology care.

In our patient, therapy was not only multifaceted but also frequently modified to optimize control of main risk factors. In the last decade, in fact, a constant control of hypertension, anemia, glycemia and proteinuria was associated with remarkable stability of renal function in our patient. In particular, we observed a decrement of 24-h proteinuria, both as an absolute or eGFR-standardized value. Nowadays, the reduction of proteinuria is recognized as the main predictor of better renal survival [44]. Besides BP and anemia control and anti renin-angiotensin system (anti-RAS) therapy, the long-lasting adherence to prescribed dietary salt and protein restriction, as well as the glycemic control and treatment with active vitamin D possibly contributed to the observed proteinuria reduction [45–48].

We had planned to start RRT at some point of time; two AVFs were in fact prepared though unsuccessfully. However, the adequate control of extracellular volume and metabolic parameters never prompted us to implement emergency dialysis. Therefore, we periodically faced the critical question on the timing of chronic dialysis start in this asymptomatic patient. This question remains a common one in renal clinics today, especially when nephrology follow-up is prolonged, as in the case of our patient.

Conclusions

No evidence-based indication on the correct timing of RRT start is today available. The available literature is in fact limited. Further studies are definitely needed in the Nephrology setting with more appropriate design, longer follow-up and, particularly, in larger patient populations including all age strata. Results of the ongoing European QUALity Study on treatment in advanced chronic kidney disease (EQUAL study) may probably shed some more light on this controversial issue; however, also this study will provide information limited only to patients of 65 years and older [http://www.equal-study.org/it/].

Our systematic analysis and case report suggest that close monitoring and multifactorial care by the nephrologist may allow to maintain ‘pre-dialysis’ renal function over the long term independently of age. While awaiting for new solid evidence, the decision on dialysis start should be personalized, taking into account the global clinical and metabolic picture rather than being dictated only by specific abnormalities or fixed GFR thresholds [2, 4]. Notably, the careful attempts to maintain ESRD patients dialysis-free should be pursued not only for the sake of improving prognosis and quality of life of patients but also because the economic resources today available for health care are limited. Therefore, postponing RRT, even for a few months, may also translate into major savings for the health system dialysis budget [37, 49, 50]. This “saved” money could be wisely used to potentiate renal clinics dedicated to the prevention of ESRD.