Estimating GFR Based on Cystatin C Versus Serum Creatinine in Patients With Cancer

Appropriate dosing of renally cleared medications relies on accurate assessment of estimated glomerular filtration rate (eGFR). The most commonly used method of evaluating GFR in both clinical practice and research is measurement of serum creatinine.

According to Paul E. Hanna, MD, and colleagues, despite increasing precision of eGFR equations, serum creatinine-based eGFR (eGFR_cr) remains inaccurate and can overestimate eGFR, particularly in patients with sarcopenia. The overestimation may result in inaccurate dosing of mediations that require adjustment based on eGFR, including antibiotics, muscle relaxants, antiepileptic drugs, blood thinners, and antiarrhythmic medications.

To test the hypothesis that patients with cancer would have substantially lower eGFR based on cystatin C level (eGFR_cys) than based on eGFR_cr, the researchers recently conducted a cohort study. Results were reported in JAMA Network Open.

The cohort included adult patients with cancer at two major academic cancer centers in Boston, Massachusetts. Eligible patients had creatinine and cystatin C measured on the same day between May 2010 and January 2022. The baseline date was the date of the first simultaneous eGFR_cr and eGFR_cys measurement.

The primary study exposure was eGFR discordance, defined as an eGFR_cys that was more than 30% lower than the eGFR_cr. The primary outcome of interest was the risk of specified medication-related adverse events within 90 days of the baseline date: (1) supratherapeutic vancomycin trough level greater than 30 µg/mL; (2) trimethoprim-sulfamethoxazole-related hyperkalemia (>5.5 mEq/L); (3) baclofen toxic effect; and (4) supratherapeutic digoxin level (>2.0 ng/mL). The secondary outcome, assessed using a multivariable Cox proportional hazards regression model, was 30-day survival of participants with versus without eGFR discordance.

A total of 1869 patients met inclusion criteria and were included in the analysis. Mean age was 66 years, 49% (n=921) were female, 51% (n=948) were male, and 80% (n=1486) identified as non-Hispanic White.

Of the overall cohort, 29% (n=543) had an eGFR_cys that was more than 30% lower than the eGFR_cr. The reference group included patients with concordant eGFR. In a multivariable logistic model, the factors associated with an eGFR_cys that was more than 30% lower than the eGFR_cr included White race (adjusted odds ratio [aOR], 1.43; 95% CI, 1.05-1.95; P=.03), cirrhosis (aOR, 1.68; 95% CI, 1.03-2.77; P=.04), use of a diuretic (aOR, 1.60; 95% CI, 1.15-2.24; P=.005), recent corticosteroid use (aOR, 1.65; 95% CI, 1.23-2.21; P<.001), hypoalbuminemia, defined as serum albumin level <3.0 g/dL (aOR, 6.09; 95% CI, 4.16-8.98; P<.001), anemia, defined as hemoglobin level <10.0 g/dL (aOR, 1.98; 95% CI, 1.38-2.83; P<.001), and eGFR_cr-cys (aOR, 0.99; 95% CI, 0.98-1.00; P<.001).

The most common baseline factors associated with an eGFR_cys that was more than 30% lower than the eGFR_cr were hypoalbuminemia and anemia.

Patients whose eGFR_cys was more than 30% higher than the eGFR_cr (n=195; 10% of the cohort) were younger, had fewer medical comorbidities, were less likely to smoke, and had higher baseline eGFR_cr-cys than patients with concordant eGFR. Due to the frequent use of antiretroviral medications that inhibit secretion of creatinine, HIV infection was more common in patients with an eGFR_cys that was more than 30% higher than the eGFR_cr.

In the multivariable logistic model, the factors associated with an eGFR_cys that was more than 30% higher that the eGFR_cr were age (aOR per 10-year increase, 0.75; 95% CI, 0.66-0.86; P<.001), White race (aOR, 0.62; 95% CI, 0.43-0.89; P=.01), diabetes (aOR, 0.60; 95% CI, 0.40-0.89; P=.01), and eGFR_cr-cys (aOR, 0.99; 95% CI, 0.99-1.00; P=.03).

A total of 268 patients who reported receiving vancomycin within 90 days of the baseline date had their vancomycin trough level measured. Patients with an eGFR_cys that was more than 30% lower than the eGFR_cr were more likely to have significantly elevated vancomycin trough levels than patients with concordant eGFR and patients with an eGFR_cys that was more than 30% higher than eGFR_cr (43 of 179 [74%) vs 7 of 77 [9%]; P=.01).

Of the 235 patients who received trimethoprim-sulfamethoxazole within 90 days of the baseline date and had their serum potassium level checked within 30 days, grade 2 hyperkalemia was numerically higher in those with an eGFR_cys that was more than 30% lower than the eGFR_cr compared with the eGFR-concordant group.

Thirty-one patients received a prescription for baclofen within 90 days of baseline. Five of the 19 patients with an eGFR_cys that was more than 30% lower than the eGFR_cr developed clinical evidence of baclofen toxic effects, prompting discontinuation of the medication compared with none of the patients with concordant eGFR or with an eGFR_cys that was more than 30% higher than the eGFR_cr.

Ninety-nine patients received a prescription for digoxin. Of those 24 had an eGFR_cys that was more than 30% lower that the eGFR_cr. Seven of the 24 had a digoxin trough level greater than the therapeutic range compared with zero of the 10 patients in the eGFR-concordant group (P=.08).

Seven percent of the cohort (n=126) died within 30 days and 9% (n=160) were lost to follow-up prior to 30 days. The 30-day mortality rate was significantly higher in patients with an eGFR_cys that was more than 30% lower than the eGFR_cr compared with patients with concordant eGFR. Following adjustment for age, sex, race and ethnicity, baseline comorbidities, laboratory studies, and medication use, patients with an eGFR_cys that was more than 30% lower that the eGFR_cr had a 1.98-fold increased hazard of death within 30 days (95% CI, 1.26-3.11; P=.003). There was no increased risk of death in patients whose eGFR_cys was more than 30% higher than the eGFR_cr compared with patients with concordant eGFR.

Limitations cited by the authors included using a one-time assessment of serum creatinine and cystatin C, and the inability to identify cancer stage and measures of functional status from the electronic health record.

In conclusion, the researchers said, “In this cohort study, we found that an eGFR_cys that was more than 30% lower than the eGFR_cr was associated with increased renally cleared medication-related adverse events. Future prospective studies are needed to improve and personalize the approach to GFR estimation and medication dosing in patients with cancer.”

Takeaway Points

1. In patients with cancer, estimation of glomerular filtration rate (GFR) based on serum creatinine (eGFR_cr) may overestimate the GFR.
2. Researchers reported results of a study testing the hypothesis that having an eGFR based on cystatin C (eGFR_cys) that is substantially lower than an eGFR_cr would be common in patients with cancer.
3. Among patients with cancer in this study, supratherapeutic drug levels and medication-related adverse events were more common in those with eGFR_cys more than 30% lower than their eGFR_cr.

Source: JAMA Network Open