Shattered renal tablets of stone

The bibles of nephrology separate acute renal failure (ARF) from chronic renal failure (CRF) as definitely as the old testament from the new.

But this separation lies among the remnants of the stone tablets of nephrology, called into question in the New England Journal of Medicine a year ago. Chawla et al¹ suggested half the renal commandments are no longer credible – why?

Old orthodoxy

For several decades, the core tenets of nephrology have included:

• ARF and CRF renal failure are separate conditions
• ARF does not predispose to CRF
• CRF does not lead to ARF except that decompensation may superimpose “acute on chronic” renal failure
• the factors initiating CRF differ from those underlying its self-sustaining linear progression – that is, glomerular filtration rate (GFR) falls at a consistent rate
• the usual outcome of ARF is recovery of normal (or pre-existing) renal function
• the fall of GFR in CRF results from loss of functional nephrons. In ARF it can result from damaged nephrons, but not necessarily – it can also result from inappropriate redistribution of renal perfusion, away from the cortex (hence the glomeruli) towards the medulla

Chawla and Kimmel² regard these as factoids; instead we should consider an integrated syndrome of diminished GFR with acute and chronic stages and outcomes determined by factors that include the balance between adaptive and maladaptive responses. Accordingly, patients that have recovered from ARF should be followed up by a nephrologist. Only 12% are monitored long-term, unlike survivors of myocardial infarction (76%). Distrust of the prevailing wisdom has mushroomed in the past 10 years. Some of the reasons follow.

Certainty of transience

During the 1980s nephrologists regarded the factors initiating CRF as separate from those sustaining its progression; the latter were far more important.

First, CRF is asymptomatic until well after 50% of nephrons have been lost – whatever initiated CRF happened months or years before diagnosis. But if self-sustaining linear progression was separate, the key factors were probably still operating and amenable to research.

Second, control of progression might avoid the need for dialysis and transplantation. Patients would be protected from end-stage renal disease (ESRD). Problematically, evidence concerning progression rested on plasma creatinine – moreover, progression was non-linear in a third of patients³.

Even 15 years ago, it began to look likely CRF was actually the outcome of a cumulative succession of separate renal insults over a lifetime, including the toxic side effects of medications, rather than a single process⁴. That would make it almost impossible to model authentically in laboratory animals; like most multifactorial diseases, a spontaneous model from veterinary medicine would probably be more fruitful.

Chipping at foundations

The doubts raised by Chawla et al1 are more fundamental. They argue the distinction between ARF and CRF is false; they are manifestations of a single clinical complex.

The emphatic importance attached to linear progression as a feature of CRF is also crumbling. Some progression is linear, some not, and some CRF does not progress. Individuals may switch between these during their illness – a key clinical priority is to distinguish between progressors, non-progressors and improvers⁵.

Chawla and Kimmel2 cited evidence showing strong association of ARF with subsequent CRF, also with progression of pre-existing CRF and a long-term increase in the risk of ESRD. The results in some studies are alarming – an eight-fold to 28-fold increased risk of advanced CRF following ARF. The course of renal disease following ARF depends on the extent of the fall in GFR, the reversibility of the precipitating insult and the balance between effective and maladaptive repair responses. Repetitive insults worsen long-term outcomes.

The “natural” age-related decline of GFR makes elderly patients more vulnerable to adverse drug reactions and potentially to ARF, as does reduced GFR due to congestive heart failure. Moreover, patients suffering reduced renal function following cardiac surgery were at risk of subsequent ESRD. While the emphasis is on changes in renal function, both ARF and CRF subject patients to the adverse effects of uraemia, including those on their kidneys.

Chawla and Kimmel suggested the severity, not the duration of ARF, is decisive; beyond a threshold, ARF initiates CRF. He also summarised multiple mechanisms underlying the link between ARF and CRF, including some already implicated in progression of CRF.

Chawla and Kimmel concluded: “Separation of ARF and CRF into distinct syndromes as taught in medical school is not in the best interest of an integrated approach to the long-term care of patients with kidney disease. Rather, we propose consideration of the state of diminution in GFR as a clinical entity, which has differential initiation and expression in time and along the spectrum of magnitude of GFR. Phases of disease include initiation, repair and long-term outcomes.”

Chawla and Kimmel are supported by Palevsky⁶, who added that even small increases of plasma creatinine profoundly affect mortality, hence emphasising acute renal injury rather than “failure”.

Data from Bucaloiu et al⁷ are particularly disturbing: 1,600 ARF patients, carefully matched with 3,650 appropriate controls, had twice the risk of subsequent CRF and twice the mortality.

Moreover, children with ARF, unlikely to have pre-existing CRF, have an increased risk of developing it and, among surgical patients without pre-existing CRF, those who recover from ARF eventually have increased mortality⁸.

The most extreme position comes from Venkatchalam et al⁹, who elaborated “a comprehensive hypothesis regarding the pathophysiology of acute kidney injury as it relates to the progression of kidney disease”, but rather than a unitary hypothesis they present a dazzling array of possibilities.

Too good to be true?

A loud blast on the trumpet does not inevitably mean the walls of Jericho will fall. An article by Rifkin et al¹⁰ did not specifically cite Chawla, but did raise serious counterarguments and highlighted weaknesses and gaps in the revolutionary case.

Undoubtedly, while most cases of ARF are reversible, some are associated with subsequent appearance of CRF. What they doubt is whether this is cause and effect. They cited Bradford Hill’s rules for attributing associations to a causal link; these are worth repeating:

• causes must precede effects
• greater exposure increases incidence or magnitude
• the association can be replicated using different methods in different settings
• the association is consistent with known biological or pathological processes (or existing beliefs need reassessment)
• the putative effect can be altered by an experimental intervention
• a single cause produces the effect without other changes
• the association is consistent with the natural history of the disease or laboratory findings
• the effect of similar factors may be considered in other populations or circumstances
• stronger association may make causality more likely

The danger of assuming causality is exemplified by anaemia and CRF. The association with poor prognosis is real, but not causal; correction does not improve outcomes.

Balancing probabilities

The problem for both sides of the debate is ARF and CRF share many risk factors – it is hard to prove ARF precedes CRF rather than simply occurring during its subclinical stages.

Moreover, some risk factors, having caused ARF, may persist and, subsequently, they, rather than ARF, may cause CRF.

Perhaps hardest to explain, if ARF causes CRF, is the welcome fact renal donors do not suffer CRF despite their sudden drastic reduction of GFR – also the fact patients requiring dialysis for severe ARF usually recover normal renal function.

The debate is also beset by technical problems – for example, dependence on retrospective analysis of  routine plasma creatinine measurements, rather than actual changes in GFR.

Creatinine also reflects muscle mass and dehydration, both affected by disease, including renal disease; plasma creatinine and its derivatives poorly reflect actual changes in GFR^11,12.

Furthermore, outpatient diagnoses of ARF can easily be decompensated CRF; less than 10% of ARF cases are referred to a nephrologist for their follow-up, hence the risk that ARF occurring during pre-existing CRF will be wrongly regarded as its cause.

The dogma is in the dock; the agnostics are at the gates. But the debate remains unresolved. Whether it causes CRF, ARF kills more patients than breast or prostate cancer, heart failure or diabetes combined.

The United States, Canada, Europe and Australia have about 1.5 million survivors of ARF annually, but 15% to 20% will develop CRF within two years^2,13. This is not simply an academic spat – the verdict will affect treatment and patient outcomes.