Abstract
This study characterizes the exclusion of patients with chronic kidney disease (CKD) in randomized clinical trials of drug treatment for bladder, breast, colorectal, lung, and prostate cancer.
Chronic kidney disease (CKD) is common in patients with cancer, with reported prevalence ranging from 12% to 53% at cancer diagnosis.1 Cancer patients with concomitant CKD have worse cancer outcomes than those with normal kidney function.2 Despite the prevalence of kidney dysfunction and its prognostic implications, there is limited evidence to guide cancer treatment in patients with CKD. One reason may be the exclusion of CKD patients from clinical trials (as has been documented in cardiovascular studies).3 We sought to quantify and characterize the exclusion of patients with CKD in randomized clinical trials of anticancer drugs.
Methods
We performed a systematic search of MEDLINE for randomized trials of drugs for the 5 most common solid cancers (bladder, breast, colorectal, lung, and prostate) in 6 high-profile general medicine and oncology journals (selected based on impact factor, and listed in Table 1) from January 2012 through December 2017. We excluded trials of surgery, radiation, and supportive care. Nonrandomized studies and pooled analyses were also excluded.
Table 1. Characteristics of Randomized Clinical Trials of Anticancer Drugs Examined for the Exclusion of Patients With Chronic Kidney Disease.
| Trials, No. (%) | Patients, No. | Trials Explicitly Excluding Kidney Disease, No. (%) | P Valuea | |
|---|---|---|---|---|
| Overall | 310 (100) | 282 889 | 264 (85) | |
| Publication, yb | .16 | |||
| 2012 | 58 (19) | 46 567 | 50 (86) | |
| 2013 | 60 (19) | 73 745 | 55 (92) | |
| 2014 | 46 (15) | 35 264 | 37 (80) | |
| 2015 | 47 (15) | 37 437 | 45 (96) | |
| 2016 | 47 (15) | 41 486 | 32 (68) | |
| 2017 | 52 (17) | 48 390 | 45 (87) | |
| Enrollment start, y | .24 | |||
| 1995-2000 | 12 (4) | 32 018 | 8 (67) | |
| 2001-2006 | 91 (29) | 122 789 | 77 (85) | |
| 2007-2012 | 176 (57) | 113 061 | 154 (88) | |
| 2013-2017 | 31 (10) | 15 021 | 25 (81) | |
| Trial enrollment, No. of patients | .52 | |||
| <100 | 14 (5) | 1016 | 11 (79) | |
| 100-200 | 35 (11) | 5282 | 29 (83) | |
| 201-500 | 85 (27) | 28 269 | 77 (91) | |
| 501-1000 | 89 (29) | 61 893 | 75 (84) | |
| >1000 | 87 (28) | 186 429 | 72 (83) | |
| Cancer type | .45 | |||
| Bladder | 4 (1) | 959 | 3 (75) | |
| Breast | 111 (36) | 144 052 | 87 (78) | |
| Colorectal | 52 (17) | 42 619 | 48 (92) | |
| Lung | 96 (31) | 50 175 | 86 (90) | |
| Prostate | 47 (15) | 45 084 | 40 (85) | |
| Intervention type | .02 | |||
| Chemotherapy | 78 (25) | 60 986 | 68 (87) | |
| Biologic or immunotherapy | 87 (28) | 81 802 | 78 (90) | |
| Endocrine therapy | 31 (10) | 65 331 | 18 (58) | |
| Targeted agents | 84 (27) | 43 725 | 78 (86) | |
| Other therapy | 30 (10) | 31 045 | 28 (93) | |
| Trial phase | .82 | |||
| 2 | 55 (18) | 11 094 | 45 (82) | |
| 2/3 | 7 (2) | 7610 | 7 (100) | |
| 3 | 246 (79) | 263 735 | 210 (85) | |
| 4 | 2 (1) | 440 | 2 (100) | |
| Funding source | .61 | |||
| Industry | 208 (67) | 168 941 | 177 (85) | |
| Government | 39 (13) | 32 634 | 34 (87) | |
| Both | 63 (20) | 81 314 | 53 (84) | |
| Journal | .09 | |||
| JAMA | 4 (1) | 6287 | 3 (75) | |
| Journal of Clinical Oncology | 137 (44) | 113 495 | 124 (91) | |
| Journal of the National Cancer Institute | 5 (2) | 4786 | 4 (80) | |
| Lancet | 16 (5) | 36 465 | 12 (75) | |
| Lancet Oncology | 112 (36) | 80 233 | 90 (80) | |
| New England Journal of Medicine | 36 (12) | 41 623 | 31 (86) |
The P values presented relate to the effect estimates from the multivariable logistic regression model.
The 5-year period prior to initiation of the systematic review was initially selected as the time frame of interest (to assess trials of recent cancer therapies, including novel immunotherapies and targeted agents). The review was subsequently updated to include trials published up to December 31, 2017.
Two reviewers independently assessed published articles, protocols, and supplementary materials for each trial to determine whether patients with CKD were excluded. Corresponding authors were contacted in cases of ambiguity regarding exclusions. Multivariable logistic regression was used to assess for trial characteristics associated with exclusion of CKD. Analyses were performed using SAS (SAS Institute), version 9.4. Two-sided P values less than .05 were considered statistically significant.
Results
A total of 1599 citations were obtained from the MEDLINE search, of which 1198 were excluded based on title and abstract screening, yielding 401 articles for detailed review. Of these, 310 articles met eligibility criteria and were included in this study.
Eighty-five percent of trials excluded patients with CKD (Table 1). Trials of endocrine-based therapies (vs other interventions) were less likely to exclude CKD patients. None of the other measured trial characteristics were associated with CKD exclusion.
Serum creatinine threshold values were the most common exclusion criteria (62%), followed by creatinine clearance (CrCl; 44%) (Table 2). Few trials used estimated glomerular filtration rate (eGFR) thresholds for exclusion (5%). Multiple exclusion criteria pertaining to kidney function were reported in 90 trials (34%).
Table 2. Thresholds Used for Exclusion of Patients With Kidney Disease in Randomized Clinical Trials of Anticancer Drugs (N = 264 Trials)a.
| Measurement for Kidney Function–Based Exclusiona | No. of Trials (%)b |
|---|---|
| Serum creatinine value | 162 (62) |
| Serum creatinine value relative to ULN | 129 (49) |
| >ULN | 16 (6) |
| >1.25-times ULN | 7 (3) |
| >1.5-times ULN | 93 (35) |
| >2-times ULN | 6 (2) |
| >2.5-times ULN | 6 (2) |
| >5-times ULN | 1 (0.4) |
| Absolute serum creatinine value, mg/dL | 33 (13) |
| >1.5 | 17 (6) |
| >2.0 | 15 (6) |
| >4.0 | 1 (0.4) |
| CrCl, mL/min | 115 (44) |
| <60 | 38 (14) |
| <50 | 44 (17) |
| <45 | 10 (4) |
| <40 | 12 (5) |
| <30 | 11 (4) |
| eGFR, mL/min/1.73 m2 | 14 (5) |
| <60 | 5 (2) |
| <50 | 4 (2) |
| <45 | 1 (0.4) |
| <30 | 4 (2) |
| Proteinuria | 31 (12) |
| Nonspecified renal exclusionc | 41 (16) |
| Multiple exclusion criteria related to kidney functiond | 90 (34) |
Abbreviations: CrCl, creatinine clearance; eGFR, estimated glomerular filtration rate; ULN, upper limit of normal.
Some included trials utilized multiple measures of kidney function for exclusion.
The percentages reported reflect a denominator of 264 (the total No. of trials with exclusions based on kidney function).
An example of nonspecified renal exclusion is “adequate” or “intact” kidney function.
An example of multiple exclusion criteria related to kidney function is serum creatinine, CrCl, and proteinuria.
Among trials using CrCl or eGFR exclusion criteria, thresholds were usually above 45 mL/min, which reflects mild or mild to moderate CKD (in total, 39% of trials excluding CKD). Severe CKD thresholds (CrCl or eGFR <40 mL/min) were infrequently used as exclusion criteria (11%).
Discussion
Eighty-five percent of recent trials of therapies for the 5 most common malignancies (published in selected high–impact factor journals) excluded patients with CKD. This proportion exceeds that observed in cardiovascular trials published from 1985 through 2005 (56%).3 This finding is concerning because it was estimated that 32% of deaths among patients with CKD in 2005-2009 were attributable to malignancy.4 As a result of trial underrepresentation, patients with CKD may not be considered for cancer therapies that have potential to improve morbidity and mortality.
Most trials used serum creatinine or CrCl thresholds to exclude patients, despite data demonstrating that these are suboptimal measures of kidney function in cancer patients.5 Given the availability of more accurate and validated methods for estimating kidney function (eg, eGFR formulas), use of serum creatinine alone is inadequate.
The exclusion of patients based on kidney function is appropriate when CKD is severe and prognosis-limiting or when concerns exist regarding potential nephrotoxicity or adverse events due to bioaccumulation of renally cleared drugs. However, this review suggests that patients with CKD with only mild to moderate kidney dysfunction are often excluded, and frequently from trials of interventions (eg, biologics or immunotherapies) for which there may be no pharmacologic basis for renal exclusions.6
Limitations of this study include the restrictions to common cancers and to articles in high–impact factor journals, which may have led to exclusion of negative or smaller trials that included patients with CKD. Also, the number of potential patients with CKD that were excluded could not be determined, as none of the studies assessed reported these data.
The exclusion of patients with CKD should be based on appropriate measures of kidney function and justifiable clinical or pharmacokinetic rationale. Judicious broadening of eligibility criteria in cancer trials to include the growing population of patients with CKD may enable more patients to benefit from novel cancer therapies while balancing the potential risk of adverse events.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.Launay-Vacher V, Oudard S, Janus N, et al. . Prevalence of renal insufficiency in cancer patients and implications for anticancer drug management. Cancer. 2007;110(6):1376-1384. doi: 10.1002/cncr.22904 [DOI] [PubMed] [Google Scholar]
- 2.Iff S, Craig JC, Turner R, et al. . Reduced estimated GFR and cancer mortality. Am J Kidney Dis. 2014;63(1):23-30. doi: 10.1053/j.ajkd.2013.07.008 [DOI] [PubMed] [Google Scholar]
- 3.Coca SG, Krumholz HM, Garg AX, Parikh CR. Underrepresentation of renal disease in randomized controlled trials of cardiovascular disease. JAMA. 2006;296(11):1377-1384. doi: 10.1001/jama.296.11.1377 [DOI] [PubMed] [Google Scholar]
- 4.Navaneethan SD, Schold JD, Arrigain S, Jolly SE, Nally JV Jr. Cause-specific deaths in non-dialysis-dependent CKD. J Am Soc Nephrol. 2015;26(10):2512-2520. doi: 10.1681/ASN.2014101034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Torres da Costa E Silva V, Costalonga EC, Coelho FO, Caires RA, Burdmann EA. Assessment of kidney function in patients with cancer. Adv Chronic Kidney Dis. 2018;25(1):49-56. doi: 10.1053/j.ackd.2017.10.010 [DOI] [PubMed] [Google Scholar]
- 6.Ryman JT, Meibohm B. Pharmacokinetics of monoclonal antibodies. CPT Pharmacometrics Syst Pharmacol. 2017;6(9):576-588. doi: 10.1002/psp4.12224 [DOI] [PMC free article] [PubMed] [Google Scholar]