Understanding Risks and Our Responsibility to Living Donors

Nearly 70 years after the first living donor kidney transplant, the practice of living donation continues despite incomplete information about the long-term risks of donation and the absence of resources for lifelong postdonation care. As members of the kidney community, we have a responsibility to understand the inherent limitations of existing data sources that inform donor risk and to advocate for a new system that will provide better information to future donors, maintain donor kidney health, and enable the safe expansion of living donation.

Our understanding of long-term donor risks is challenged by the inherent limitations of uncontrolled observational studies; by the infrequency of events, necessitating large data sources to estimate risk; by the relatively short available follow-up; and by the complexities to derive a valid control population. Most of the information about long-term donor outcomes is limited to death and kidney failure obtained by linking data from transplant registries to national death, dialysis, and waiting list registries.^1–3 These studies provide reliable estimates of the absolute risk of death and kidney failure in donors, and they show that these outcomes are rare (50 in 10,000 donors develop kidney failure during 20 years of postdonation observation) and that events vary by donor age, sex, ancestry, recipient relationship, place of residence, and predonation clinical characteristics, including kidney function and body mass index.³

Studies to determine the relative risk of donation compared with controls contrast donor outcomes recorded in registries with the outcomes of control subjects enrolled in studies such as NHANES in the United States or the HUNT study in Finland.^1,2 These studies rely on the assumption of conditional ignorability to draw causal claims from observational data. Under this assumption, the assignment of a subject to the control or treatment group can be assumed to be random when conditioned on observable characteristics of the study subjects, and missing data can be treated as occurring at random so that the effects of unobserved factors and missing data can be ignored. This assumption is invalid in most living donor studies. Living donors are either biologically related and/or share lifestyle and environmental exposures with their recipients, which may be risk factors for kidney disease. Underlying comorbid conditions, psychosocial factors, behavioral attributes, and environmental exposures are likely systematically different between donors and controls derived from unrelated epidemiologic studies, leading to biases and imprecision that are particularly relevant when comparing cohorts with low event rates. Additional concerns with attempts to determine the relative risk of donation in retrospective analyses include the use of noncontemporaneous controls, leading to differences in outcome ascertainment and era effects, and the necessity for complicated matching procedures that allow the same donor to contribute multiple events to the estimation of donor event rates, resulting in inflated relative risk estimates. Finally, many of the risks (such as donor age) are nonlinear, and thus, small differences in baseline characteristics between donors and controls may lead to erroneous conclusions if one assumes a linear risk during follow-up.^4,5

Prospective studies to determine relative risk estimates of adverse health outcomes in donors compared with sibling controls, which are more likely to meet the ignorability assumption, were attempted but were simply too small to be informative.^6,7 In the current era, a small but rigorous prospective, controlled study reported that donors continued to show a small increase in kidney function after 9 years of donation compared with control subjects who had the expected age-related decline in kidney function, but it found no difference in BP, urine protein, urine albumin, glucose, hemoglobin A1c, insulin, and lipoproteins between donors and controls.⁸

Estimation of the lifetime risks of kidney failure by extrapolation of pooled data in nondonors with CKD risk factors may be useful in calculating the predonation lifetime risk of kidney failure in potential donors with similar characteristics, including low-normal eGFR, hypertension, and high-normal urine albumin-creatinine ratios.⁹ However, these models do not permit estimation of postdonation risk, and in addition, the model predicting lifetime risk was on the basis of studies in which the longest follow-up was only 16 years. Therefore, these lifetime predonation risk estimates are less useful in younger donors in whom such estimates are most relevant to inform decision making. Lifetime risk estimates of the combined outcome of eGFR<30 ml/min or kidney failure are also available using data from donors, but these estimates are also limited by the average duration of follow-up of only 17 years.¹⁰ The available observational data have allowed some understanding of the heterogeneity of donor risk, but our ability to combine demographic and clinical factors to determine more precise absolute risk estimates in donor subgroups in the form of a heat map is still limited to two factors (i.e., donor age and ancestry) due to the small number of events in subgroups.³

A key concept to be considered with all studies is that of attributable risk: that is, the incremental risk of the intervention (donation) over baseline risk. This has been largely overlooked in previous analyses. The risk of donation can only be assessed if one knows the baseline (without donation) relative risk of any confounder being studied. For instance, the risk of kidney failure in obese persons is higher than in nonobese persons. Therefore, a finding that obese donors have increased risk for kidney failure postdonation does not necessarily mean that donation contributed to this difference in outcome because the higher incidence may be the result of the inherent risks of obesity. In fact, if obese donors had the same outcomes as nonobese donors, that would be quite remarkable given the robust evidence of the risks of obesity in the general population. We would posit that this is one of the key concepts to be considered. What is the risk of donation over one’s absolute baseline risk? It has been extraordinarily difficult to derive control groups that are similar enough to donors to assess the specific change of baseline risk specifically attributable to donation. If attributable risk is simplified to the difference in donation absolute risk and nondonation absolute risk, then we must ensure that both sides of this equation are truly assessing equivalent populations.

Whether the long-term risk of adverse health outcomes in donors with certain demographic or clinical considerations can be mitigated through judicious donor selection as suggested in the recent study of obese donors by Ibrahim et al.¹¹ published in this edition of JASN remains an open question given the limited duration of follow-up in this study. The duration of follow-up is a paramount consideration because the incidence of native kidney disease increases with age, and it should not be surprising that the incidence of kidney failure in donors also increases exponentially with longer postdonation observation.³ The practical implication of these observations is that the acceptance of donors with predonation risk factors may be best limited to donors of older age in whom the duration of follow-up data is sufficient to support donation in the context of a model of shared decision making.

Given the reality of incomplete understanding about the long-term health outcomes of donors, what strategies will enable the responsible expansion of living donation? Investment in long-term prospective studies with appropriate controls that not only capture the sentinel events of death and kidney failure but also, intermediate health outcomes, including cardiovascular disease, CKD, and psychologic outcomes, should be supported.¹² Investment in such efforts will not inform clinical practice for decades but is nonetheless critical to improve risk assessment for the next generation of donors. Similarly, studies to determine the implications of genetic factors for kidney disease, such as the ApoL1 gene, for living donation must be supported. Evaluation of such novel factors in donors will not only inform living donor risk but may also accelerate understanding of the mechanisms by which novel factors contribute to kidney disease.

Importantly, studies to determine the cause of kidney failure in donors consistently implicate the role of new postdonation health events, suggesting the importance of postdonation health maintenance.¹³ The obligation to provide postdonation care was emphasized in Dr. Joseph Murray’s Nobel Lecture. Dr. Murray described how prior to the donor surgery for the first successful living donor transplant, Ronald Herrick, the donor for his twin brother Richard, asked whether the hospital would be responsible for his health care for the rest of his life if he decided to donate his kidney. Dr. Hartwell Harrison, the surgeon for the donor, said, “Of course not.” However, he immediately followed with the question, “Ronald, do you think anyone in this room would ever refuse to take care of you if you needed help?”¹⁴ Numerous commentators have decried the lack of financial support for long-term postdonation care on moral grounds, but there has been no movement to fund donor follow-up in the United States beyond the current 2 years mandated by the Organ Procurement and Transplant Network. Given the realities of finite health care resources, policy makers are unlikely to act on principal alone. By preventing adverse health outcomes in donors, facilitating the collection of much needed information about donor health, and allowing more potential donors with predonation health considerations to be accepted for donation, funding a program to ensure donor health maintenance will likely be cost effective, if not cost saving.¹⁵ We have mined our existing data sources to their useful limit; as a community, it is our responsibility to advocate for dedicated resources to ensure better postdonation care and data acquisition for living donors.

Disclosures

J.S. Gill reports research funding from Astellas; honoraria from Takeda; and scientific advisor or membership with the American Society of Transplantation and the Declaration of Istanbul Custodial Group. J. Schold reports consultancy agreements with Guidry and East, Novartis, Sanofi Corporation, and Transplant Management Group; honoraria from Novartis and Sanofi Inc; scientific advisor or membership as a data safety monitoring board member for Bristol Myers Squibb and Nephrosant; and speakers bureau with Sanofi. The remaining author has nothing to disclose.

Funding

None.