3.2. Cause-Specific Graft Loss for All Patients Combined and Stratified by Recipient Age
A cause-specific cumulative hazard plot for all patients combined is presented in
Figure 1A. Other than NFG, which only occurred in four patients during the first 2 weeks post-transplant, the hazard rates of GFNA, Other GF, and DWFG (i.e., slopes of the cumulative hazard curves) were reasonably constant over time beyond 5 years post-transplant (with 22, 31, and 40 patients experiencing these cause-specific graft losses, respectively). Thus, there did not appear to be any falloff in these cause-specific hazard rates over time, even at 18 years post-transplant.
Cause-specific cumulative hazard plots for patients stratified by recipient age (at transplant) <50 (N = 82) vs. ≥50 (N = 68) years are presented in
Figure 1B,C, respectively. Here, one can see that the cause-specific patterns of failure were dramatically different between the younger and older patients. First, GFNA and Other GF were the primary causes of graft loss for patients <50 years of age at transplant, particularly during the first 10 years post-transplant, whereas DWFG was the primary cause of graft loss for patients ≥50 years of age at transplant. The observed proportions of graft loss due to GFNA, Other GF, and DWFG for patients <50 years of age at transplant were 21/82, 21/82, and 8/82, respectively, in comparison with 1/68, 10/68, and 32/68 for patients ≥50 years of age at transplant, respectively. Thus, the hazard rates of developing GFNA and Other GF were distinctly higher for patients <50 years of age at transplant, and the log-rank test comparing death-censored graft survival (combining NFG, GFNA, and Other GF as one outcome) between recipient age <50 and ≥50 years yielded
p = 0.002. Conversely, the hazard rate of DWFG was distinctly higher for patients ≥50 years of age, with the log-rank test comparing the hazard rate of DWFG between recipient age <50 and ≥50 years yielding
p < 0.000001. Second, the hazard rate of developing GFNA was clearly higher for patients <50 years of age at transplant in comparison with patients ≥50 years of age (
p = 0.0005), with GFNA occurring in only one patient ≥50 years of age. Third,
Figure 1B shows that in patients <50 years of age, the hazard rates of GFNA and Other GF were of similar magnitude (and both higher when compared with their respective hazard rates in
Figure 1C); thus, even if GFNA were to become eliminated as a cause of graft loss, the younger patients (compared with older patients) would remain at higher risk of developing Other GF over time.
CIF estimates (±SE) of the percentages experiencing cause-specific graft loss for all patients combined and then stratified by recipient age <50 vs. ≥50 years appear in
Table 2 and
Figure 2A–C, respectively. For all patients combined, estimates at 216 months (18 years) post-transplant for NFG, GFNA, Other GF, and DWFG were 2.7% ± 1.3%, 16.9% ± 3.4%, 23.8% ± 3.8%, and 30.2% ± 4.1%, respectively. Of note, among the 22 patients who experienced GFNA, only 13.6% (3/22) occurred during the first 36 months post-transplant. For patients <50 years of age at transplant, estimates at 216 months (18 years) post-transplant for NFG, GFNA, Other GF, and DWFG were 3.7% ± 2.1%, 28.8% ± 5.4%, 29.3% ± 5.4%, and 11.3% ± 3.8%, respectively. In comparison, for patients ≥50 years of age at transplant, estimates at 216 months (18 years) post-transplant for NFG, GFNA, Other GF, and DWFG were 1.5% ± 1.5%, 1.7% ± 1.7%, 15.1% ± 4.4%, and 55.1% ± 7.2%, respectively. The log-rank test for a difference in the CIF for GFNA between recipient age <50 and ≥50 years was highly significant, yielding
p = 0.00002. While the log-rank test for a difference in the CIF for Other GF between recipient age <50 vs. ≥50 years was not statistically significant, yielding
p = 0.13, the log-rank test for a difference in the CIF for DWFG between recipient age <50 and ≥50 years was highly significant, yielding
p < 0.000001.
3.3. Multivariable Analyses of GFNA
Results of the Cox stepwise regression analyses for the hazard rates of developing GFNA and the CIF for GFNA (i.e., subdistribution hazard) were essentially the same in that two factors were jointly associated with a significantly higher GFNA rate: younger recipient age (as a continuous variable) (
p = 0.0001 for the GFNA hazard rate;
p = 0.00003 for the GFNA subdistribution hazard) and African-American or Hispanic recipient (
p = 0.0007 for the GFNA hazard rate;
p = 0.001 for the GFNA subdistribution hazard) (Cox model coefficients and corresponding SEs are shown in
Table 3). For both Cox models, no other baseline variable was significant in either univariable or multivariable analysis, including the treatment arm of maintenance immunosuppression in which the patient had been assigned (results not shown).
The multivariable influence of recipient age (categorized by <35, 35–49, and ≥50 years) and race/ethnicity (African-American or Hispanic vs. non-African-American and non-Hispanic) are reflected by the cumulative hazard plots and CIF estimates shown in
Figure 3A,B, respectively. CIF estimates (±SE) of the percentages experiencing GFNA stratified by these two baseline variables are also shown in
Table 4. The two higher risk groups were clearly patients <35 and 35–49 years of age at transplant who were either African-American or Hispanic (10/26 and 10/30 of these patients <35 and 35–49 years of age had GFNA), with estimated cumulative percentages developing GFNA (±SE) by 18 years post-transplant being 41.5% ± 10.2% and 36.9% ± 9.6%, respectively. For the other four groups defined by recipient age and race/ethnicity (2/94 of these patients combined had GFNA), the estimated cumulative percentages of developing GFNA by 18 years post-transplant ranged between 0.0 and 6.7%.
3.4. Characterization of NA Occurrence
During 18 years of post-transplant follow-up, 18.7% (28/150) of the cohort had become NA. As described above, 22/28 of these patients subsequently developed GFNA. A cumulative hazard plot of NA occurrence for all patients combined appears in
Figure 4. Of note, a large one-year spike in NA occurrence occurred during the fourth year post-transplant, where 6/28 determinations of NA occurrence were made (apparently, right after 36 months post-transplant, when Medicare insurance coverage ended for all nondisabled transplant recipients <65 years of age). Interestingly, the hazard rate of NA occurrence (slope of the cumulative hazard curve) appeared to be reasonably constant beyond 5 years post-transplant (with no evidence of any falloff). Among the 28 patients who became NA, the median time to becoming NA was 46.2 months (range: 3.3–179.6 months) post-transplant; most of these determinations, 67.9% (19/28), occurred beyond 36 months post-transplant, respectively. Among the 22 patients who subsequently experienced GFNA, the median time from becoming NA until GFNA occurrence was 20.7 months (range: 2.4–118.6 months).
Cox stepwise regression of the hazard rate of becoming NA yielded the same two unfavorable factors as those found for the GFNA hazard rate: younger recipient age (as a continuous variable) (
p = 0.000004) and African-American or Hispanic Recipient (
p = 0.00007) (Cox model coefficients and corresponding SE’s are shown in
Table 5). For this Cox model, no other baseline variable was significant in either univariable or multivariable analysis (results not shown). The observed percentages of patients who became NA, stratified by recipient age and race/ethnicity, were as follows: 42.9% (24/56) for African-American and Hispanic recipients <50 years of age, 7.7% (2/26) for non-African-American/non-Hispanic recipients <50 years of age, 6.5% (2/31) for African-American and Hispanic recipients ≥50 years of age, and 0.0% (0/37) for non-African-American/non-Hispanic recipients ≥50 years of age.
Descriptive characteristics for each of the 28 patients who became NA are shown in
Table 6; 89.3% (25/28) developed a biopsy-proven acute rejection (BPAR) and/or chronic rejection (CR) episode (with T-cell, B-cell, or both types being present) as a direct consequence of their NA behavior, which also subsequently triggered GFNA in 22 patients. Time-to-BPAR (or CR) along with the corresponding histologic grade are shown for each of these 25 patients in
Table 6; median time-to-BPAR/CR following the start of NA was 8.1 months (range: 0.6–118.2 months). Of note, the rejection episode directly due to NA behavior was the patient’s first BPAR in 15/25 cases, the patient’s second BPAR in 3/25 cases, and CR (without documented BPAR) in 7/25 cases. Histologic evidence of B-cell BPAR, i.e., acute antibody mediated rejection, was also present in 7/18 of the BPAR cases.
There were three patients who experienced rejection as a direct consequence of their NA behavior but without subsequently developing GFNA. Patient no. 10 had increasing renal dysfunction (i.e., serum creatinine) following her second BPAR (at 41.6 months), and then experienced DWFG due to a cardiovascular event at 55 months post-transplant. Patient no. 11, who admitted to having completely stopped taking her immunosuppressive medications twice, once for 4 weeks and then again for 8 weeks prior to developing a first BPAR at 103 months post-transplant, experienced a complete recovery of her renal function following treatment of the rejection episode. This patient became (and remained) reasonably compliant in taking her immunosuppressive medications thereafter and was alive with a completely functioning graft at 192 months post-transplant. Patient no. 23, following his first BPAR at 169.2 months, was alive with increasing renal dysfunction at 215 months post-transplant.
There were also three patients who became NA but without ever developing rejection as a direct consequence of their NA behavior. Patient no. 16, who became NA at 52.6 months post-transplant and exhibited NA behavior for approximately 16 months thereafter, never developed rejection and, subsequently, became (and remained) reasonably compliant in taking his immunosuppressive medications. This patient ultimately developed Other GF due to acute and chronic B-cell (antibody mediated) rejection at 164 months post-transplant. Patients no. 24 and no. 28 became NA at 120 months and 180 months post-transplant, respectively. While both patients remained intermittently NA thereafter, neither patient ever developed rejection nor suffered from any long-term renal dysfunction. These two patients were alive with completely functioning grafts at 204 months and 221 months post-transplant, respectively.
Among the 22 patients who developed GFNA, 21/22 continuously remained NA from the time of becoming NA until GFNA (see
Table 6). Patient no. 4 was briefly NA for about 2 months starting at 15.3 months post-transplant (she experienced an empirically treated AR at 16.0 months post-transplant). While this patient was subsequently compliant in taking her immunosuppressive medications for approximately 3 years, she once again became (and remained) NA for about 26 months until experiencing a first BPAR followed by GFNA at 76.6 months and 78.3 months post-transplant, respectively.
For each of the 28 NA cases, the attending physician determined that the patient had become NA in taking his/her immunosuppressive medications.
Table 6 shows that the great majority of NA patients, 89.3% (25/28), had admitted to their NA behavior. In fact, 42.9% (12/28) of the NA patients admitted to having completely stopped taking their immunosuppressive medications; median length of time in which the patient admitted to having completely stopped taking his/her immunosuppression was 4 weeks (range: 1 week–18 months). In four additional cases, it was suspected that the patient had completely stopped taking his/her immunosuppressive medications prior to presenting with the BPAR (or CR). Overall, 11/28 NA patients displayed undetectable trough levels for at least one of his/her immunosuppressive medications; low and/or undetectable immunosuppression drug trough levels were documented in 26/28 NA patients.
Lastly, indicated and/or acknowledged reasons for the patient’s NA behavior included a strictly psychological component in 7/28 cases, and a combination of both financial and psychological components in most, 71.4% (20/28), of the cases. In only one case was the reason for the patient’s NA behavior not documented. Financial reasons included the patient’s expressed inability to pay for the immunosuppressive medications, including co-payments. Among the 20 patients having a documented financial component to their NA behavior, insurance issues were a contributing financial reason in 17/20 cases (4 patients were unemployed). Among the 27 patients with a documented psychological component to their NA behavior, clinical depression was previously diagnosed in 7/27 cases, and some combination of patient apathy, anxiety, and irritability (characteristics of clinical depression) was indicated in the remaining 20/27 cases.