Can COMBINED Magnetic Resonance Imaging Measure the Progression of Kidney Disease?

CKD is a major health care problem, affecting approximately 10% of the population worldwide. Despite the development of drugs that can slow down the progression of CKD, such as renin-angiotensin blockers and more recently SGLT2 inhibitors, the number of patients with kidney failure is still increasing (1).

Some patients with CKD are perfectly stable for many years, whereas others suddenly deteriorate, leaving no time to prepare them for kidney transplantation or dialysis. Every nephrologist knows that predicting kidney function decline (mainly derived from the yearly change in creatinine-based eGFR) is a difficult task. Scores such as the kidney failure risk equation provide an estimation, but remain approximate and do not integrate the underlying disease process. A recent score developed for persons with diabetic kidney disease explained only 12% of the variability of their eGFR decline (2). Kidney biopsy is generally considered as the best tool to predict outcome, but is only performed in a minority of cases. This is partly because of fear of complications, but also because of the impossibility to stop antiaggregation drugs or patient refusal. Therefore, international guidelines recommend regular controls of eGFR in all patients with CKD to differentiate rapid from slow progressors. This strategy is time-consuming, and may result in a multitude of follow-up visits for patients who end up having a stable eGFR for many years. Also, creatinine is not a sensitive biomarker, as it only increases after considerable damage has already occurred in the kidneys.

Numerous researchers have therefore tried to find new, more sensitive biomarkers. An ideal biomarker should not only predict kidney outcome, but also provide information on the underlying disease process and its activity.

In this context, magnetic resonance imaging (MRI) is an interesting candidate. Conventional MRI provides high-resolution images of the macroscopic structure of the kidneys, which may point toward possible causes of CKD (such as autosomal dominant polycystic kidney disease). The past two decades have witnessed the arrival of new MRI techniques that can quantify functional or microscopic properties of the kidneys, such as cortical and medullary blood flow, and tissue oxygenation and fibrosis. A short, nonexhaustive overview of these so-called “functional” MRI techniques is shown in Table 1.

Table 1.

Overview of most commonly used functional MRI techniques for the kidney

MRI Technique	Basic Principle	Outcome Variable
Blood oxygenation level–dependent MRI	Tissue oxygenation is estimated by using the paramagnetic properties of deoxyHb that shortens transverse relaxation time constant T2*	R2* (1/T2*, sec−1): high R2* corresponds to high local deoxyHb levels and presumes low tissue oxygenation in kidneys and vice versa
		R2* is also influenced by magnetic field inhomogeneities (because of air in bowels), hemoglobin levels, hydration status, and salt intake
Arterial spin labeling	Subtraction technique that uses magnetically labeled water protons to measure cortical (and medullary) perfusion	Tissue blood flow of kidneys, expressed in ml/min per 100 ml
		No gold standard against which to validate. Technically challenging, motion and breathing artifacts
Diffusion-weighted imaging	The ADC of water is detected and provides information on tissue microstructure	ADC (mm2/s): in general, lower ADC indicates greater kidney fibrosis
		ADC also depends of renal blood flow and tubular flow
Phase-contrast MRI	Moving protons of blood in the renal artery induce a phase-shift that corresponds to its velocity	Renal blood flow (per artery) in ml/s
		Sensitive to background noise and aliasing; technically difficult in case of tortuous renal arteries
T1 mapping	In T1 mapping, a whole-kidney map of all quantified T1 relaxation times is obtained. Alterations of T1 are nonspecific, but may indicate interstitial edema or inflammation	Msec
		T1 values are influenced by many other factors such as body temperature, perfusion, and fibrosis of the kidney tissue

MRI, magnetic resonance imaging; deoxyHb, deoxygenated hemoglobin; R2*, inverse transverse relaxation time T2*; ADC, apparent diffusion coefficient.

Functional MRI has several advantages. First, it can be performed without intravenous contrast media, and is thus not contraindicated in patients with (advanced) CKD. Second, separate results are obtained for the left and right kidney, which is interesting in diseases with asymmetric presentation, such as renal artery stenosis or reflux. Third, the entire kidney can be analyzed, instead of a small part as in kidney biopsies.

Many research groups have performed studies to assess whether functional MRI can predict eGFR decline. Initially, single-center studies were performed that used only one functional MRI technique. As such, it was shown in several (but not all) studies that high cortical relaxation rate (R2*; corresponding to lower oxygenation) and low apparent diffusion coefficient (ADC; corresponding to more fibrosis) are associated with faster eGFR decline. In recent years, it became technically possible to perform different techniques within a single MRI session (duration ≤1 hour), and their simultaneous combination (so-called multiparametric MRI) was integrated in research protocols (3).

The study by Srivastava et al. published in this issue of CJASN (4) is among the first completed multicenter, multiparametric, functional MRI studies in CKD. The authors obtained baseline diffusion-weighted imaging and blood oxygenation level–dependent MRI for 122 participants in the COMBINE (CKD Optimal Management with Binders and NicotinamidE) trial, a randomized, double-blinded trial of nicotinamide and lanthanum carbonate versus placebo in patients with stage 3b-4 CKD. They found that lower baseline cortical ADC was associated with more rapid loss of eGFR, but this association was no longer significant after adjustment for baseline albuminuria. Cortical R2* was not associated with the change in eGFR over time.

Hence, their results were not in line with previous single-center studies (5,6). Some limitations of this study may partly explain this discrepancy. Their study was slightly underpowered, as power calculations were on the basis of primary biochemical end points, and fewer patients than planned were recruited. Further, the follow-up period was rather short (1 year) and annual eGFR decline was modest (−2.3 ml/min per 1.73 m² per year), strongly reducing the likelihood of a significant association between MRI parameters and eGFR decline.

Notwithstanding these limitations, this study clearly has strengths. It shows, for the first time, that multicenter, functional MRI studies are possible. Moreover, the study shows that most patients with CKD can undergo MRI, despite the fact that many were obese; one patient was excluded owing to claustrophobia, three patients were excluded owing to large body habitus, and six patients were excluded owing to metallic implants. Finally, although not the aim of this trial, it also shows that functional MRI can be incorporated in randomized, clinical trials, and identifies hemodynamic and structural effects of the drug of interest.

This study also illustrates some of the difficulties encountered in this field. Patient preparation is an issue, as some parameters are sensitive to external factors; for example, R2* values are influenced by dietary sodium intake, glycemia, and hydration status. ADC is on the basis of the degree of movement of water molecules in tissues; the latter is not only influenced by tissue fibrosis, but also by capillary density or tubular flow (and thus possibly by diuretics). All these factors need to be taken into account when performing functional MRI. Further, different magnetic resonance sequences are used by distinctive vendors and centers, and image analysis is not always performed in the same way. The functional MRI community is aware of these issues, and thanks to a joint effort of international experts funded by the European Union called PARENCHIMA (Magnetic Biomarkers for Chronic Kidney Disease), consensus papers on image acquisition and analysis have been recently published (7,8).

Another issue is of more methodological nature. Many MRI studies assess the association between MRI parameters and eGFR decline over a relatively short period. In cases where the parameter does not correlate with eGFR decline, one may conclude that the parameter has limited potential to predict outcome. At the same time, we know that eGFR is not a sensitive marker of kidney damage, especially in the early stages. Although this will need larger trials with longer follow-up, studies that focus on hard end points would be welcome. Trials that repeat functional MRIs are also of interest, as they are the only way to demonstrate that MRI can show kidney damage before any change in eGFR occurs. An example of such a study was recently published (9).

In conclusion, the demonstration of the feasibility of multiparametric, multicenter, functional MRI studies, such as the one by Srivastava et al. (4), and the recently achieved international consensus on how to perform and analyze functional MRI, represent important steps forward. However, more studies demonstrating the benefit (and cost-effectiveness) of functional MRI in CKD management are necessary before the integration of this promising technique in clinical practice.

Disclosures

Dr. Pruijm has nothing to disclose.

Funding

Dr. Pruijm reports grants from the Swiss National Science Foundation (FN 320030-169191).