Age And The EGFR-dependent Risk For Adverse Clinical Outcomes Ⅱ

Age and the association between eGFR and all-cause mortality Relative association (hazard ratio)

In the CKD-PC study, the curve of the adjusted hazard ratio associated with progressively lower eGFR versus a constant reference value became less steep (i.e. increased to a lesser extent) with older age [3]. The relative mortality rate ratio (eGFR 45 versus 80 mL/min/1.73 m2) decreased as age increased (Fig. 2, top right panel). Using a common reference eGFR of 80 mL/min/1.73 m2 for every age category, the adjusted hazard ratio for mortality began to increase significantly when the eGFR was approximately <60 mL/min/1.73 m2 across all age groups. This finding of relative change in “speed” was adopted to support the use of a single eGFR threshold for defining CKD [19]. However, change in speed does not inform about the distance the disease has traveled by level of exposure and modifier (eGFR and age, respectively), and thus cannot stand alone in the assessment of these associations. In other words, we cannot understand the distance traveled (risk at certain time horizons of interest) from the change in speed.

CLICK HERE TO GET CISTANCH FOR KIDNEY HEALTH

Another issue to consider when reporting relative measures is that relative associations depend on the choice of the reference category. If we agree that eGFR declines with older age, we should use an age-adapted eGFR reference category. Instead of using a common reference eGFR for all ages [3], a re-analysis of the data that changed the reference category within each age group reached apparently different results: hazard ratios for mortality increased in a similar way when the threshold of CKD was increased from 60 to 75 mL/min/1.73 m2 in individuals aged 18–54 years, maintained equal to 60 mL/min/1.73 m2 in those aged 55–64, and lowered to 45 mL/min/1.73 m2 in those aged ≥65 years [14]. When using relative measures of association, the choice of the reference category matters.

Absolute rate

In the CKD-PC study [3], the curve of the absolute mortality rates associated with lower eGFR was steeper with older age, suggesting that differences in mortality rates associated with progressively lower eGFR were larger in older than in younger age categories. For example, the difference in mortality rates between an eGFR of 45 versus 80 mL/min/1.73 m2 was 27.2 deaths per 1000 person-years for age ≥75 years and 9.0 deaths per 1000 person-years for age 18–54 years (Fig. 2, top panels). Comparing differences in mortality rates (speeds instead of distances) across age groups is problematic [14], because mortality increases with age, resulting in different baseline mortality risks across age groups. In addition, the implication of a young person (say a 40-year-old) dying in the next 10 years is very different from an old person (80-year-old) dying in the next 10 years. Finally, mortality rates are seldom constant [20].

Absolute risk 

Absolute risk data overcome the limitations of relative associations (which are reference-sensitive) or rates (interpretationand time-variability). Absolute risk data can be obtained after controlling for confounding by regression (adjusted cumulative incidence functions from cause-specific or sub-hazard regression models [9, 10]) or stratification. A Canadian population-based cohort study reported data on the underlying mortality risks in people 65 years or older with normal or mild albuminuria. In this study including 127 132 people who had CKD according to current eGFR criteria (a fixed threshold of 60 mL/min/1.73 m2), as many as 54 342 (43%) were 65 years or older and had a baseline eGFR of 45–59 mL/min/1.73 m2 with normal or mild albuminuria. The difference in the 5-year absolute risk of death between people with an eGFR of 45–59 mL/min/1.73 m2 and those with an eGFR of 60–89 mL/min/1.73 m2 (non-CKD controls) was minimal in age groups 65–74 and ≥75 years (Fig. 3, top panels). While 5-year mortality risk was higher for eGFR of 15– 44 mL/min/1.73 m2 within each age subgroup (65–69, 70–74, 75– 79 and ≥ 80 years), the difference between eGFR 45–59 and 60–89 mL/min/1.73 m2 was small in all age categories >65 years (Fig. 4, top panel). Figure 5 shows the distribution of 5-year risk by event type (death and kidney failure) over age treated as a continuous variable. The relationship between age and 5-year risk was only slightly altered by eGFR levels of 45–59 and 60–89 mL/min/1.73 m2 in elderly people with normal or mild albuminuria (the majority of elderly people with CKD according to current eGFR criteria; Fig. 4, bottom panel) [7].

Figure 2: Relative and absolute rates of death and ESKD associated with eGFR according to age categories (CKD-PC study). ESKD was defined as initiation of dialysis, kidney transplantation or death coded as due to kidney disease other than acute kidney injury. Data were from the CKD-PC study [3]. Left panels: absolute mortality (top) and ESKD rates (bottom) associated with an eGFR of 45 versus 80 mL/min/1.73 m2 across age categories. Right panels: differences and ratios of mortality rates (top) and ESKD rates (bottom) associated with the change in eGFR from 45 to 80 mL/min/1.73 m2.

failure Relative association (hazard ratio)

Data from the CKD-PC study show that the relative hazards of ESKD with lower eGFR were comparable across age categories, with the exception of a slightly stronger association in the youngest age group with eGFR of 41–51 mL/min/1.73 m2 [3]. What is the meaning of this apparent lack of evidence of interaction? As compared with a reference eGFR of 80 mL/min/1.73 m2, an eGFR level of <60 mL/min/1.73 m2 was associated with an increased hazard of ESKD in both younger and older adults [3]. Does the increased relative hazard of ESKD at eGFR <60 mL/min/1.73m2 across all age groups support the use of a single eGFR threshold for defining CKD?

Absolute rate 

In the same study, the authors reported mean incidence rates for ESKD according to eGFR within each age category [3]. Rates of ESKD were close to 0 for eGFR ≥60 mL/min/ 1.73 m2 and can be appreciated only when eGFR was below 45 mL/min/1.73 m2, particularly for older individuals: 23.0 and 9.8 cases per 1000 person-years for age 65–74 and ≥75 years, respectively [3]. The ESKD rate difference associated with an eGFR of 45 versus 80 mL/min/1.73 m2 appeared to be smaller with increasing age (for example, 45.1 versus 8.0 cases per 1000 person-years for age ≥75 versus 18–54 years; Fig. 2, bottom panels), although “the differences in absolute risk were not significant except for a limited GFR range in which the adjusted average ESKD incidence rate was higher in the youngest age group” [3]. While the proper interpretation of relative hazards requires information on absolute measures of disease occurrence, we should remember that the authors reported rates, not risks. A rate is a “speed” and thus provides only indirect information about the absolute risks over a time interval. Estimates of the absolute risk of kidney failure will help us better understand the impact of decreased eGFR on the outcome of kidney failure across age groups.

Absolute risk 

Although the relative hazard of ESKD dramatically increases with lower eGFR regardless of age, the absolute risk of kidney failure becomes smaller with older age. Considering elderly people with normal or mild albuminuria, even if the relative hazard of ESKD is higher for eGFR of 45–59 versus 60–89 mL/min/ 1.73 m2, the absolute risks of kidney failure at 5 years are very low (as low as 0.1%) and thus of questionable clinical relevance (Fig. 3, bottom panel). Conversely, the 5-year absolute risk of death is very high (9.7% and 7.3% among people 65–74 years, respectively; Fig. 3, top panel) [7]. Focusing on relative measures, we may be distracted by noticing that the 5-year risk of kidney failure is about one to two times higher in people with eGFR 45– 59 versus 60–89 mL/min/1.73 m2 up to the age of 80 years and then it may be lower after the age of 80 years (Fig. 5, right panel). However, looking at the risk range of the two events (y-axis tick labels) we see what really matters: in elderly people with normal or mild albuminuria with older age the 5-year risk of death increases from 5% to 100%, while the 5-year risk of kidney failure ranges between 0.01% and 0.1%, regardless of whether eGFR is above or below 45 mL/min/1.73 m2.
To illustrate, let us revisit the bike example. Let us assume there are two pairs of bikers, one aged 65 years and one aged 85 years. For each pair of bikers, one person has an eGFR of 45– 59 mL/min/1.73 m2 and one 60–89 mL/min/1.73 m2. The person with lower eGFR travels faster toward either event. However, the location of each pair of bikers in the disease trajectory depends on age. At any given time horizon from a time origin when eGFR and age are assessed, older pairs will have traveled a longer journey in the mortality trajectory (will be more likely to die) and a minimal distance on the trajectory toward kidney failure (will have traveled a small fraction of journey toward kidney failure). The opposite will happen to the younger pair (Fig. 6).

Of note, absolute measures are dependent on many factors, which need to be considered carefully when interpreting study findings [3, 7]. For example, the CKD-PC study estimated absolute rates for a given value of eGFR within an age category using regression models. Rate estimation (i.e. speed assessment) does not need to account for competing events. However, risks cannot be derived from rates directly in the presence of competing risks. The Canadian study reported absolute risks, including competing risks, stratified by eGFR and age. In addition, the population under study, and thus baseline risks, are different in the two studies. The Canadian study used population-based data and included people with CKD defined as sustained eGFR below a threshold for more than 3 months (following the guideline’s recommendation [21]), and its main analyses focused on elderly people with normal or mild albuminuria, for whom CKD definition is debated. The CKD-PC study included general and high cardiovascular-risk cohorts as well as CKD cohorts (whether CKD was defined as a sustained reduction of eGFR below a threshold for a period of time is unclear), and only 7.3% of the CKD-PC participants were 75 years or older. Finally, in the CKD-PC study, ESKD was not a comprehensive definition of kidney failure. The choice of avoiding kidney replacement and opt for conservative care of kidney failure is more common with older age [18].

Figure 4: Relative and absolute risks of death and kidney failure at 5 years, by baseline eGFR and age categories (65–69, 70–74, 75–79, ≥80 years) in people with normal/mild albuminuria. This figure is from Liu et al., Accounting for Age in the Definition of Chronic Kidney Disease (2021), by permission of JAMA Intern Med [7]. Risks of kidney failure (KF) and death at 5 years by index eGFR (baseline eGFR) and age in elderly people with normal/mild albuminuria.