Cystatin C: more evidence of its benefit

 I am aware of the fact that I have done many blogs on cystatin C, perhaps too many. But, it seems to me that having as accurate a measure of estimated GFR is really important in terms of patient care, since so many issues depend on that, including the potential accuracy of medication-dosing based on renal function, as well as the treatment of the renal insufficiency itself. In that light, I herewith present the latest evaluation, this one finding that incorporating Cystatin C into the calculation of eGFR in older patients improved the prediction for all-cause mortality, the necessity for renal replacement therapy, as well as 6 other clinically important outcomes (see ckd cystatin c better in older pts AIM2024 in dropbox, or doi:10.7326/M23-1138).

 

Details:

-- 82,154 people at least 65 years old who had outpatient creatinine and cystatin C measurements in this Swedish study, with information from 2010-2019

   -- the SCREAM database (Stockholm CREAtinine Measurements) includes information on demographics, healthcare utilization, laboratory tests, dispensed drugs, diagnosis, and vital status

    -- this study included only people who did not have a history of KFRT (kidney failure with replacement therapy), and did have creatinine and cystatin  measurements (they used the first set of measurements ifmultiple were done)

        -- 19% of these older adults had simultaneous creatinine and cystatin C measurements, creating a sample of a 2154 people

-- mean age 77, 55% female

-- hypertension 80%, antihypertensive meds 76%, diabetes 45%, history of cardiovascular disease 41%, total cholesterol 195/HDL 55, mean urinary albumin-creatinine ratio 17

-- mean eGFR_cr 67; eGFR_cr-cys 61, eGFR_cys 54  [eGFRcr is the  eGFR using creatinine in the new non-race based equation; eGFR_cys uses cystatin C, which never had race included; and  eGFR_cr-cys  is the equation that combines the creatinine and the cystatin C values]

       -- these calculations were based on the CKD-EPI equations 2021: see https://www.kidney.org/professionals/kdoqi/gfr_calculator

-- the primary exposures they assessed were eGFR_cr and eGFR_cr-cys , but they also included eGFR_cys

-- covariates assessed in the statistical analysis below included sex, hypertension, diabetes, history of cardiovascular disease, antihypertensive meds, total and HDL cholesterol, and urinary albumin-creatinine ratio

 

-- main outcomes (8 in all): all-cause mortality, cardiovascular mortality, KFRT, all-cause hospitalizations, infections, MI/stroke, heart failure, acute kidney injury (AKI)

 

-- median follow-up 3.9 years

 

Results:

-- the overall results are as follows in these graphs:

 

 

 

-- these outcomes were impressive in that:

    -- both the eGFR_cr-cys  and eGFR_cys  curves were associated with higher hazard and incidence rate ratios for all of the outcomes,  as opposed to the curve for eGFR_cr  

    -- both of these curves were linear, finding a continuous relationship between higher adverse outcomes with decreasing renal function; in contrast, the curve for eGFR_cr  was U-shaped and had  worse outcomes at the higher eGFR values, a counterintuitive finding

 

For example, they provided the following adjusted hazard ratios (adjusted for the co-variates noted above) and incidence rate ratios for individuals with 60 vs 80 mL/min/1.73 m² for eGFR_cr, eGFR_cr-cys, and eGFR_cys respectively:

-- all-cause mortality: 1.0 (0.9 to 1.0), 1.2 (1.1 to 1.3), and 1.3 (1.2 to 1.4)

-- cardiovascular mortality: 1.0 (0.9 to 1.1), 1.3 (1.2 to 1.4), and 1.4 (1.2 to 1.6)

-- KFRT: 1.4 (0.7 to 2.8) 2.6 (1.2 to 5.8), and 1.5 (0.6 to 3.9)

-- heart failure: 1.2 (1.1 to 1.4), 1.5 (1.4 to 1.7), and 1.7 (1.6 to 1.9)

-- AKI: 1.6 (1.4 to 1.9), 2.3 (2.0 to 2.6), and 2.3 (1.9 to 2.7)

    -- ie, eGFR_cys  did poorly as compared to the other 2 measures

    -- there was not much difference between adverse outcomes using either eGFR_cr-cys, and eGFR_cys, and there were overlapping confidence intervals for all of them in these calculations

    -- similar divergences were found for those comparing eGFR levels 30 or 45 versus 80

 

-- in comparing those aged 70 vs 80, all event rates were higher in the 80 year olds, except for KFRT and AKI (KFRT was actually higher in the 70 year olds and they were about the same for AKI)

-- Also, absolute incidence rates for all outcomes were higher with eGFR_cr-cys

    -- and they were higher depending on the eGFR. For someone who was 70yo:  eGFRcr-cys levels of 30, 60, and 80 were associated with recurrent hospitalizations of 65 per 100 person-yrs (61 to 70), 44 per 100 person-yrs (43 to 46), and 39 per 100 person-yrs (38 to 41)

 

-- using eGFR_cr-cys instead of  eGFR_cr resulted in 31.2% of older persons to be reclassified correctly per actual observed outcomes, and mostly to be at higher risk, and this was increasingly true with increasing age

--similar findings were true for subgroup analyses by sex, cardiovascular disease, and diabetes, in those with UACR below 30, or by using the European Kidney Function Consortium equations

 

Commentary:

-- kidney function decreases with age, such that 23% to 44% of those at least 65 years old have an eGFR_cr <60 mL/min/1.73 m², the level at which adverse outcomes start to increase, including kidney failure and all-cause mortality. This GFR level is associated with more adverse outcomes in younger vs older people

    -- the concern regarding age was part of the inspiration for this study, since this was based on eGFR_cr and might not be as predictive of adverse consequences as one involving cystatin C

-- as mentioned in prior blogs, there are real, serious concerns about large gaps between both the creatinine- and cystatin-based estimated GFR as well as their combination vs the actually measured GFR, with a recent article finding huge divergence in both under- and over-estimations that often crossed over more than 1 CKD stage: https://gmodestmedblogs.blogspot.com/2022/07/egfr-not-such-great-estimate-of-renal.html

    -- But, that being said, a prior study from the SCREAM database (the one used in this current study) involving younger patients, found that creatinine was not a predictor of either future adverse cardiovascular or renal outcomes (the current analysis evaluated more outcomes), and that in general there was a pretty linear relationship between the eGFR_cys minus the  eGFR_cr value: the greater the difference when eGFR_cys was lower than eGFR_cr the higher the adverse event risk. And the eGFR_cr added nothing to the predictive model: https://gmodestmedblogs.blogspot.com/2023/12/cystatin-c-better-predictor-of-bad.html

-- though cystatin-based eGFR measurements are pretty clearly more predictive of adverse outcomes than creatinine-based measurements, it is important to note the concerns with both of them:

    -- creatinine levels are influenced by muscle mass, diet, a pretty long list of many other medications (including salicylates, trimethoprim cimetidine, corticosteroids, vitamin D, dolutegravir, ...) and physical activity; whereas cystatin C is influenced by inflammation, obesity, smoking, thyroid diseases, and glucocorticoid use 

    -- there is an argument that since both creatinine and cystatin are affected by different non-renal causes, that the combo embodied in the eGFR_cr-cys  would be more accurate. I am not so sure that makes sense on an individual basis though it may on a community level

    -- there was one study comparing eGFR_cr, eGFR_cr-cys, and eGFR_cys , finding that eGFR_cr-cys was in fact better than the other two in eliminating race-based algorithms and tracked better with the measured GFR, especially in Black individuals (the eGFR_cys  was very close to eGFR_cr-cys, more so than the eGFR_cr  for all individuals): https://www.nejm.org/doi/full/10.1056/NEJMoa2102953 .

-- it was also interesting to note another study using cystatin-based measurements: https://gmodestmedblogs.blogspot.com/2024/03/limitations-of-nt-probnp-with-renal.html found that NT-proBNP measured for heart failure is not so accurate a clinical measure when patients have CKD, but the ratio of NT-proBNP/cystatin was much better (this study did not assess creatinine, but did confirm a profound benefit of cystatin in assessing NT-proBNP levels)

 

--this study did find that in people over 65 years old, eGFR_cr-cys had much stronger associations with clinical outcomes than those with eGFR_cr

    -- in fact, those with eGFR_cr measurements actually had a U-shaped curve, not found with the other measurements where there was a linear curve (which more appropriately fits reality)

    -- of note, the cystatin-based calculations of eGFR never had a “correction” for Black individuals (adding the race component had very little effect on the cystatin-based eGFR calculation)

-- one very impressive part of this study is that they assessed 8 different outcomes, finding a quite consistent association between eGFR_cr-cys  with all of the clinical outcomes

-- although this study did not set out to assess eGFR_cys  in particular, it was pretty clear from the graphs that this was pretty much as good as the combination eGFR_cr-cys

-- it was quite notable that almost 1/3 of the patients in this age group would be correctly reclassified in terms of their array of clinical outcomes by using a cystatin-based eGFR as opposed to a creatinine-based eGFR

 

Of note, in Sweden cystatin C is widely used to assess renal function, and the cost is only $3/test.  In our inconsistent system of health care in the US, the cost is higher and quite variable depending on insurance, and seems to varies between $99 and $595/test (https://www.findlabtest.com/lab-test/kidney-function-test/cystatin-c-labcorp-121251 )

-- both the National Kidney Foundation and the American Society of Nephrology do recommend using cystatin C  to estimate GFR, based on the fact that it seems to more accurately reflect the actual measured GFR

  

Limitations:

-- this study formally compared eGFR_cr-cys to eGFR_cr  and not eGFR_cys , though their graphs do suggest that simple measurement of eGFR based on cystatin C did seem quite accurate in terms of the 8 clinical outcomes assessed

-- this was an observational study and not a clinical trial, so conclusions reflect an association between these different measurements of the estimated GFR and clinical conditions, and cannot determine causality

 – they did not have calculations of these different eGFRs over time, only a baseline value. And it is possible that there were some significant and important variations in these values over time that might undercut the accuracy and generalizability of their results

-- there were no measured GFR values to use as the gold standard (though this is very hard to do in such a large study)

-- there was no adjustment for those items that might have influenced the creatinine and cystatin C measurements, such as muscle mass, diet, obesity status/BMI, smoking, inflammation, and drug-drug interactions

-- as with all data-mining studies, the accuracy of the output depends on the accuracy of the input; for example they rely on accurate coding of outcomes in the database that may be so accurate

-- there may be a selection bias as to which of these older individuals in the large database actually had a simultaneous creatinine and cystatin C measurement at baseline, given that only 19% of those >65yo did have them simultaneously. Why did some clinicians order both tests vs one or the other? Were these patients sicker? did they have different underlying comorbidities? were their clinicians more attuned to more current medical issues and then treated their patients differently (older studies going back 20+ years ago touted the better value of cystatin C)? Since we have no information about treatments after the initial baseline data, we cannot answer these questions. and the results of this study may not be so applicable to the general population.

 

So,

-- yet another analysis finding that cystatin C is a better indicator of an array of adverse subsequent clinical outcomes than creatinine, with perhaps some greater benefit by using the combination eGFR including creatinine over just cystatin C

-- my anecdotal (but pretty large) assessments of both creatinine and cystatin C over several years confirms a quite large divergence of results in eGFR, with pretty much everyone having either much lower eGFR_cys than eGFR_cr or similar values

    -- and this is not confined to older folks with evidently low muscle mass (where eGFR_cr will be much lower than eGFR_cys), or younger folks with evidently high muscle mass (where the opposite is likely to be true)

    -- I am finding this discordance pretty often even in people who have no apparent reason to assume that one measurement would be better than another, and this discordance can be as much as a 20-30 point lower value for eGFR_cys than eGFR_cys

 -- as mentioned above, I am also not sure that eGFR_cr-cys  is much better than eGFR_cys on an individual patient basis.

    -- the graphs and data above suggest that on a community-basis, eGFR_cys is within the same confidence intervals as eGFR_cr-cys, though this was not a primary outcome of this study and is therefore less statistically rigorous

-- my sense, therefore, is that using a cystatin C-based eGFR is the way to go, though the combined cystatin-creatinine eGFR may be a bit better

-- in my practice, I do measure both creatinine and cystatin C on pretty much everyone, and use the creatinine-based value in the future renal function assessments only if the 2 values are very close to each other