Improved Late Graft Survival and Half-Lives in Pediatric Kidney Transplantation: A Single Center Experience

Authors

  • Abhay Vats,

    Corresponding author
    1. a Department of Pediatrics, Children's Hospital of Pittsburgh, USADepartments of bSurgery and cPediatrics, University of Minnesota, USA
    Search for more papers by this author
  • a , Kristen Gillingham,

    1. a Department of Pediatrics, Children's Hospital of Pittsburgh, USADepartments of bSurgery and cPediatrics, University of Minnesota, USA
    Search for more papers by this author
  • b Arthur Matas,

    1. a Department of Pediatrics, Children's Hospital of Pittsburgh, USADepartments of bSurgery and cPediatrics, University of Minnesota, USA
    Search for more papers by this author
  • and b Blanche Chavers c

    1. a Department of Pediatrics, Children's Hospital of Pittsburgh, USADepartments of bSurgery and cPediatrics, University of Minnesota, USA
    Search for more papers by this author

Abstract

We evaluated variables associated with improved late graft survival in 290 children transplanted between 11/1/1984 and 12/31/1997, and who had > 1 year graft survival. We studied the following variables: age, gender, race, primary disease (diseases prone to recurrence, i.e. hemolytic uremic syndrome, focal segmental glomerulosclerosis or oxalosis vs. others), primary vs. re-transplant; donor source, acute tubular necrosis, acute rejection episodes in the first year, transplant era and discharge serum creatinine. Graft half-life was defined as the time taken for 1/2 of the grafts functioning at 1 year to fail. There were 205 living donor and 85 cadaveric transplant. The cumulative graft survival at 5 and 10 years was 88% and 75% for living donor, and 72% and 46% for cadaveric, respectively. Multivariate analyses showed a higher late graft survival to be associated with: no acute rejection episodes (risk ratio 0.16, p = 0.0001), age 2–5 years (risk ratio 0.24, p = 0.0007), living donor (risk ratio 0.46, p = 0.017), primary nonrecurrent disease (risk ratio 0.29, p = 0.001), Caucasian race (risk ratio 0.40, p = 0.006). A high half-life was seen with living donor transplant (21.3 years) and the age group 2–5 years (27.5 years). Further, living donor patients with no acute rejection episodes had the best half-life of 37.6 years, while children with hemolytic uremic syndrome, focal segmental glomerulosclerosis or oxalosis had the lowest overall half-life of 5.6 years. This study finds that living donor, no acute rejection episodes, age 2–5 years, Caucasian race and having a disease not prone to recurrence are strong predictors of late graft survival. Hence, preferential use of living donor and prevention of acute rejection episodes in the first year are key variables that can improve long-term renal graft survival in children.

Introduction

Renal transplant (Tx) in children has shown steady improvement in both technique and graft survival outcome over the last two decades (1–4). However, the most significant improvement in graft survival has been generally limited to the first post-transplant (postTx) year, and rates steadily decline thereafter (1,3,4). According to the 2001 North American Pediatric Renal Transplant Cooperative Study (NAPRTCS) data, the 1-year graft survival rate is now 91% for living donor (LD) and 83% for cadaveric donor (CAD) Tx, but the rates decline significantly to 80% and 65% at 5 years for the LD and CAD groups, respectively (1). Although chronic rejection remains the major cause of graft failure, the factors which are predictive of improved graft survival beyond the first postTx year have not been the focus of recent reports and there is only limited information available on half-lives of pediatric renal allografts (1,4–7). Recently, an analysis of the United Network for Organ Sharing (UNOS) data on adult renal transplant recipients showed a significant improvement in graft half-lives (T1/2) over an 8-year study period, from 1988 to 1996 (8). The reported half-lives for pediatric renal Tx, however, have shown marked variability, ranging from 3.8 years for black teenagers to 18 years for infants (less than 2 years) in one study (4). Knowledge of strong predictors of late graft survival and variables associated with improved graft half-lives can be beneficial in planning the renal replacement therapy in children and counseling of patients and parents regarding long-term outcomes of transplantation. We present the results of analysis of variables associated with improved long-term graft survival in a large single center.

Materials and Methods

A total of 329 renal transplants were performed in pediatric (< 18 years of age) patients between 11/01/84 and 12/31/97 at the University of Minnesota Hospitals and Clinics, of which 39 grafts were lost in the first year. This study analyzed data on the 290 transplanted children who had more than 1 year of graft function. No patients were lost to follow-up. These data were acquired from analyses of a prospectively and continuously updated database maintained by full-time staff devoted to data monitoring and entry in the Department of Surgery at the University of Minnesota. The period of follow-up was till 12/31/98. The median duration of follow-up was 88 months (range 14–157 months). The patient characteristics and underlying diagnoses are shown in Table 1. Living unrelated donors comprised only 2.5% of the total (5 out of 205) and were not analyzed separately. They were combined with living related donors into the living donors category.

Table 1. : Patient characteristics
CharacteristicNumber of patients
(n = 290)
Percent
  • *

    The values in parenthesis refer to number and percent of primary renal transplants performed in each donor category.

  • **

    Diseases prone to recurrence: FSGS, HUS, oxalosis.

Transplants
 Primary22878.6
 Retransplant 6221.4
  Living donor 21 
  Cadaveric 41 
Donor source
 Living*205 (181)*70.7 (88.4)*
 Cadaveric* 85 (43)*29.3 (50.7)*
Age (in years)
 0–1 6723.1
 2–5 6221.4
 6–12 8529.3
 13–18 7626.2
Race
 Caucasian24183.1
 North American native 11 3.8
 Black 13 4.5
 Hispanic  9 3.1
 Others 16 5.5
Gender
 Male18262.8
 Female10837.2
Diagnosis
 Obstructive uropathy/PUV 6422.7
 Congenital anomalies 53 8.5
 Glomerulonephritis 23 8.2
 Polycystic, medullary cystic 19 6.7
 FSGS** 17 6.1
 Reflux nephropathy 14 5.0
 IgA, HSP  9 3.2
 HUS**  4 1.4
 Alports  4 1.4
 Oxalosis**  3 1.1
 Others 5619.3

Immunosuppression

All patients received 3 preTx random donor blood transfusions (9). The initial immunosuppression regimen consisted of prednisone (P), cyclosporine (CsA), azathioprine (AZA), mycophenolate mofetil (MMF), anti-thymocyte globulin (ATG), and Minnesota antilymphocyte globulin (ALG). Prednisone was started at 2 mg/kg/day on the day of Tx, then tapered to 0.5 mg/kg/day at 1 month and to 0.25–0.3 mg/kg/day at 1 year postTx. Azathioprine was started on the day of Tx at 5 mg/kg/day and then tapered to 2.0–2.5 mg/kg/day over the first week postTx. Cyclosporine was started at 5 mg/kg/day between postTx day 10–12 if the serum creatinine level was < 1.0 mg/dL (88.4 mol/L). If a patient developed an episode of acute rejection, CsA levels were then monitored and trough levels were maintained at around 100 ng/mL as measured by high pressure liquid chromatography (HPLC). In June 1993, we changed our CsA protocol for new Tx recipients to begin on the 2nd to 5th postTx day at 6–10 mg/kg/day if the serum creatinine was < 2.0 mg/dL (176.8 μmol/L). Patients were then monitored to maintain trough levels at 175–200 ng/mL for the first 3 months postTx. Minnesota ALG was used at 20 mg/kg/day for the first 14 days postTx from 1984 to 1992. In 1992, we changed to ATG at 20 mg/kg/day for the first 14 days postTx. Four patients received MMF instead of AZA at 15 mg/kg/dose given twice per day. The following combinations were used: CsA + AZA + P + ALG (52.1%), CsA + AZA + P + ATG (34.5%), ALG + AZA + P (4.5%), AZA + CsA + P (3.4%), MMF + P + CsA + ATG (1.4%); other combinations (4%). There was no variability in the combinations used based on donor source; preTx cytotoxic antibody levels, or underlying disease. MMF was used in four of the reTx recipients.

Acute rejection

All acute rejection episodes (ARE) were biopsy proven and classified into mild, moderate, or severe based on tubulo-interstitial and vascular involvement as previously described (10,11). A kidney biopsy was performed if patients developed a ≥ 25% elevation in baseline serum creatinine level, unexplained fever, or worsening blood pressure. No protocol biopsies were performed. No diagnosis of acute rejection was made without a biopsy and all the biopsies were read by a pathologist.

Statistical analysis

Using multivariate Cox regression analyses the following variables were studied for graft survival: age at Tx (0–1; 2–5; 6–12; 13–18), recipient sex, recipient race, primary disease [hemolytic uremic syndrome (HUS), focal segmental glomerulosclerosis (FSGS) and oxalosis vs. other], Tx number (primary vs. reTx); donor source (LD vs. CAD), acute tubular necrosis (ATN), ARE (0, 1, > 1) in the first year postTx, CsA protocol era (era 1: 11/84–5/93; and era 2: 6/93–12/97), and high or low initial discharge serum creatinine (SCr). We defined graft loss as return to dialysis, reTx, or death with a functioning graft. ATN was defined as the need for dialysis in the first week postTx. The initial discharge SCr was classified as high or low based on the age of the patient (> 5 years of age, SCr of < 1.5 mg/dL and for children < 5 years, SCr of < 1.0 mg/dL was considered low). Cox regression models included all variables being considered, and no selection procedures were used. Log–log survival plots were used to check the validity of the proportional hazards assumptions. Univariate (Kaplan-Meier) graft survival analyses were performed by the Lifetest procedure for the variables found to be significant in the Cox regression analysis. This was done to better illustrate the magnitude of the effects of these variables. The graft half-life (T1/2) was calculated by the following formula, and represents the time it takes for 1/2 of the grafts functioning at 1 year to fail (4): T1/2 = 4ln2 / (lnS1 − 1nS5).

S1 and S5 are predicted survival at 1 and 5 years, respectively, and ln is natural logarithm. Variability in half-lives among different groups was assessed by computing the weighted sum of squares of deviations from the weighted average half-life, where weights were reciprocals of squares of corresponding standard errors. These weighted sums of squares were then compared to chi-squared distributions (degrees of freedom were number of groups minus one) to obtain p-values. Statistical analyses were performed using SAS/STAT software, version 6.12 (SAS Institute, Inc., Cary, NC USA).

Results

Patient demographics

A total of 329 patients were transplanted in the study period and 39 (12%) lost their grafts (including deaths) in the first year. There were 290 pediatric renal Tx (228 primary) whose grafts survived the first year and who were analyzed. Two hundred and five children (71%) had LD Tx and 85 children (29%) had CAD Tx (Table 1). One hundred and four renal Tx were performed from 11/84 to 5/93 and 186 from 6/93 to 12/97. Primary Tx comprised 88% of the LD Tx and 51% of the CAD Tx. The majority (83%) of the children were Caucasian. The 6–12 years age group (with 29% of the children) was the single largest group. Twenty-four (9%) patients had diseases that have a potential for recurrence (i.e. HUS, FSGS and oxalosis).

Graft and patient survival

There were nine deaths (3%) in the first year postTx and nine (3%) additional deaths beyond the first year postTx. The overall patient survival rates at 1, 3, 5, and 10 years were 97%, 96%, 94%, and 92%, respectively. Graft survival rates are shown in Table 2. The overall cumulative graft survival rates at 2, 5, and 10 years were 95%, 83% and 65%. For the LD Tx group the graft survival rates at the same intervals were 97%, 88% and 75%, whereas for CAD Tx they were 90%, 72% and 46%, respectively. The cumulative 10-year graft survival for LD and CAD Tx patients is shown in Figure 1. The best graft survival rates (at years 2, 5 and 10, respectively) were seen in patients (both LD and CAD) who had no ARE in the first year [CAD: 100, 89 and 72; LD: 100, 94 and 82 (all in percentage)]. The worst long-term graft survival was seen in CAD Tx patients who had multiple (> 1) ARE (78%, 57%, and 35% at 3, 5 and 10 years, respectively). Figure 2 shows graft survival rates by number of acute rejection episodes during the first 10 years postTx in both CAD and LD groups combined.

Table 2. : Cumulative graft survival rates
CategoriesGraft survival (%) in
postTx years
p-value*
 2510
  1. LD = Living donor, CAD = Cadaver, Tx = Transplant. *p-values represent the overall values obtained for the various subgroups.

Overall 958365 
Donor source   0.0002
 LD 978875 
 CAD 907246 
Donor source and rejection episodes
 LD   0.0001
 0 Rejection1009482 
 1 Rejection 969283 
 > 1 Rejection 876942 
 CAD   0.01
 0 Rejection1008972 
 1 Rejection 967638 
 > 1 Rejecton 785735 
Age groups (years)   0.02
 0–1 979175 
 2–5 989181 
 6–12 958361 
 13–18 897166 
Disease categories   0.0008
 Non-recurrent 968667 
 Recurrent 824848 
Race   0.04
 Caucasian 978567 
 Non-Caucasian 857657 
Figure 1.

Ten-year graft survival in pediatric renal transplant recipients by donor source. Living donor vs. cadaveric donors (p = 0.0002).

Figure 2.

Ten-year graft survival in pediatric renal transplant recipients by number of acute rejection episodes for both CAD (cadaver) and LD (living donor) groups combined (ARE = 0, 1, > 1) (p = 0.0001).

Graft half-lives

Half-life data and associated variables are shown in Table 3. Since T1/2 estimation is not reasonable for subgroups with very small numbers of patients, values for some of the subgroups are not shown in Table 3. The overall incidence of acute rejection episode in the first year postTx was 52% (> 1 ARE: 25%; a single ARE: 27%). The overall T1/2 was higher in patients with no ARE (32.6 years) compared with patients experiencing one (17.6 years) or more (6.4 years) ARE (p = 0.0003, Table 3). Similar findings were seen when patients were evaluated by donor source and rejection episodes (Table 3). Children with HUS, FSGS or oxalosis had an overall T1/2 of 5.6 years vs. 16.7 years for other diseases (p = 0.00009). The lowest T1/2 in any of the subgroups was that of 2.2 years in adolescents with diseases prone to recurrence and > 1 ARE in the first year postTx. The T1/2 was higher for LD (21.3 years, p = 0.0014) and for children in the age group 2–5 years (27.5 years, p = 0.006, Table 3). There was no significant difference in T1/2 in the two CsA immunosuppression protocol eras: 15.4 years in era 1 vs. 14.9 years in era 2. Similarly, there were no differences in the T1/2 in patients discharged with high or low serum creatinine or with primary or reTx.

Table 3. : Half-lives associated with different variables
VariableGraft half-lives (in years ± SEM)p-value
  • LD = Living donor, CAD = Cadaver, Tx = Transplant.

  • *

    Data for patients with recurrent disease not shown as the number of patients in individuals groups was too low(4 or less). p = Not significant for the whole group of nonrecurrent disease.

  • **

    Data not shown as there was only 1 graft loss in these categories with the resulting T1/2 and SEM being same numbers. SEM = Standard error of mean.

Age at transplant
 0–1 (n = 67)20.3 ± 5.60.006
 2–5 (n = 62)27.5 ± 8.7 
 6–12 (n = 85)13.3 ± 3.0 
 13–18 (n = 76) 9.8 ± 1.9 
Primary diagnosis
 Non-recurrent disease (n = 259)16.7 ± 2.2 
 Recurrent disease (n = 23) 5.6 ± 1.80.00009
Donor source
 LD (n = 205)21.3 ± 3.50.0014
 Cadaver (n = 85) 9.0 ± 1.5 
Acute rejection episodes in the first year
 0 (n = 138)32.6 ± 8.10.0003
 1 (n = 80)17.6 ± 4.1 
 > 1 (n = 72) 6.4 ± 1.1 
Donor source and acute rejection episodes
 LD Tx with rejection episodes
  0 (n = 110)37.6 ± 11.30.001
  1 (n = 55)27.7 ± 9.8
  > 1 (n = 40) 7.6 ± 1.9 
 CAD Tx with rejection episodes
  0 (n = 28)21.4 ± 9.60.04
  1 (n = 25) 9.5 ± 3.0
  > 1 (n = 32) 5.4 ± 1.2 
Race
 Caucasian (n = 241)16.9 ± 2.30.01
 Non-Caucasian (n = 49) 9.3 ± 2.4 
Donor, age, and rejection in non-recurrent disease (p-NS)*
 CAD Tx < 13 years with rejection episodes
  0 (n = 18)22.7 ± 13.1 
  1 (n = 14)12.5 ± 6.3 
  > 1 (n = 21) 6.5 ± 1.9 
 CAD Tx ≥ 13 years with rejection episodes
  0 (n = 8) **  
  1 (n = 7) 4.2 ± 1.7 
  > 1 (n = 7) 4.7 ± 2.3
 LD Tx < 13 years with rejection episodes
  0 (n = 81)50.9 ± 20.8 
  1 (n = 37)25.1 ± 10.3 
  > 1 (n = 22)10.3 ± 3.6 
 LD Tx > 13 years with rejection episodes
  0 (n = 20)24.0 ± 13.9 
  1 (n = 14) **  
  > 1 (n = 10) 6.3 ± 2.8 

Multivariate analysis

Multivariate analysis showed that increased graft survival beyond the first year postTx was associated with the following variables: no ARE [risk ratio (RR) 0.16, p = 0.0001], age of 2–5 years at Tx (RR 0.24, p = 0.0007), primary disease other than HUS, FSGS or oxalosis (RR 0.29, p = 0.0014), LD donor source (RR 0.46, p = 0.017), and Caucasian recipient race (RR 0.40, p = 0.006). There was no independent effect of CsA protocol era [era 1 (before 6/93) vs. era 2 (after 6/93)]. Also, the following variables were not independent predictors of graft survival: primary vs. reTx; recipient sex, and discharge SCr. Hence, having no rejection episode, and a primary disease that is not prone to recurrence, and belonging to a younger age group (2–5 years) at the time of Tx were the significant variables associated with increased long-term graft survival.

Causes of graft losses

A total of 39 grafts (12% of 329) were lost in the first year postTx and a total of 70 grafts (24%) were lost in the study period amongst the 290 Tx patients analyzed. The major causes of graft loss in the first year were: technical/vascular complications (n = 9; 22%); recurrent disease (n = 8; 20%, including 4 with acute rejection); acute rejection (n = 8; 20%); chronic rejection (n = 6, 15%); and death with functioning graft (n = 5; 12%). The causes of the 70 graft losses beyond the first year are shown in Table 4. Chronic rejection was the single major cause of graft loss beyond the first year in both the LD and CAD groups. Of the 55 cases of chronic rejection, 48 were biopsy proven cases and 7 grafts (3 in the LD and 4 in the CAD group) were considered to be lost due to chronic rejection based on clinical diagnosis. The CAD group had a significantly higher proportion of graft losses [vs. LD group (p = 0.0055)] beyond the first year postTx. There were six deaths with functioning grafts. Known cessation of immunosuppressive therapy was the cause of graft loss in four (1.4%) patients in the period beyond the first year.

Table 4.  Causes of graft loss beyond 1st year post-transplant
 LDCADTotal
  • LD = Living donor, CAD = Cadaver.

  • a

    Numbers in parentheses were non-biopsy-proven diagnosis.

  • *

    Missing frequency = 1.

Chronic rejectiona 28 (3)27 (4) 55 (7)
Immunosuppression discontinued  2 2  4
Recurrent disease  3 1  4
Death with graft function  3 3  6
Functioning16951*220*
Total20584*289*

Discussion

This study analyzed the factors associated with improved graft survival in 290 children beyond the first year postTx at the University of Minnesota. Although a limitation of this study is that it is retrospective and from a single center, it is one of the largest single-center reports on long-term graft outcomes in pediatric renal Tx patients. We found, that in children as in adults (8,12), having no ARE in the first year postTx had the best risk ratio for long-term graft survival and was associated with the maximum half-life of the graft. The deleterious effect of increasing numbers of ARE on T1/2 was seen in both CAD and LD Tx recipients. We found that reTx showed no significant effect on long-term graft survival. Similarly, ATN was not an independent predictor of long-term graft survival.

Acute rejection is an important predictor of chronic rejection (1,5–9). In this study we found that having no ARE in the first year postTx to be the strongest independent predictor of graft survival. Further analysis of the effect of rejection on graft survival showed that within the CAD group the T1/2 is more severely affected by even a single ARE [10.7 years (with 1 rejection) vs. 25.7 years (with none)] compared to the LD group [T1/2 of 24.4 years with 1 rejection, vs. 31.9 years (with none)]. Similar results have been reported in adult renal Tx recipients. In an analysis of UNOS data on 93 934 adult patients transplanted between 1988 and 1996, Hariharan et al. reported an increase in T1/2 from 11.9 years, for patients with ARE, to 27.1 years for those who did not have any ARE (8). Although the effect of acute rejection on graft half-lives in children has not been studied, in an analysis of UNOS data on 2418 LD pediatric renal transplants, Ishitani et al. reported a 9% improvement in 7-year graft survival in those patients who did not develop acute rejection before initial discharge from the hospital (13). In our study almost half of the patients with rejection had more than one ARE. Salvatierra et al. also reported that a lower incidence of acute rejection was associated with higher 3-year graft survival rates but did not evaluate for half-lives (7). Two recent studies on long-term graft survival in pediatric renal Tx patients also did not evaluate acute rejection as an independent risk factor for graft survival (4,6).

Donor source has also been shown to have an important effect on graft survival (1,4,6–8,14). Our data show that the T1/2 of LD grafts (14.7 years) is significantly better than the T1/2 of CAD grafts (9.3 years), and these results are similar to those reported by others (4,6). Cecka et al., in an analyses of 8922 pediatric renal Tx in the UNOS database, reported graft half-lives by age with values ranging from 10.5 to 15 years for LD and 6 to 11 years for CAD Tx (4). In comparison to a single-center report on 206 pediatric renal transplants between 1970 and 1993, we found higher graft half-lives for LD patients (21.3 vs. 11.2 years) (6). This difference may reflect the larger number of LD, primary, and CsA-treated patients in our study. These results suggest that preferential use of LD sources and strategies to avoid acute rejection are important considerations in pediatric renal transplantation.

The role of age of the recipient on the graft outcome has been a subject of several prior studies (4,6,13–15). Our current study found that the age group of 2–5-year-olds had better outcome than the other age-groups overall, with respect to both graft-survival rates and graft half-lives. This is a group where the immunological response to a grafted organ may be less than in a younger-aged group, and where compliance with medication is generally under parental control (2,16,17). Graft half-life in 2–5-year-olds was almost three times that of the adolescent (13–18 years) age-group. Cecka et al. also found poorer graft half-life in adolescent (13–21 years of age) LD recipients (10.5 years) compared to that of 3–12-year-olds (15.5 years) (4).

While we found higher graft half-lives for Caucasians compared with non-Caucasian patients, there were too few non-Caucasians in this study to analyze for possible beneficial outcomes in Hispanic or Asian groups. Others have shown the detrimental effect of African-American race on graft survival (13). Kashtan et al. in an analysis of NAPRTCS data, also found FSGS to be associated with a high risk of graft failure (18). We found primary diseases that are prone to recurrence (FSGS, HUS and oxalosis) to be independently associated with short graft half-lives. While primary disease cannot be influenced, we found acute rejection was associated with significantly lower graft half lives in patients prone to disease recurrence (T1/2, 2.2 years).

In contrast to other reports, we did not find a significant difference in long-term graft survival by transplant number, i.e. primary vs. reTx (p = 0.7) (4,6). Theoretically, these patients would represent two polarized groups with respect to sensitization. The reTx patients comprised a significant proportion (21%) of the whole group and would likely have had variable degrees of prior sensitization, although we did not evaluate for cytotoxic antibody levels. Thus, one of the surprising findings of our study was that retransplantation did not confer any additional risk to long-term graft survival. In contrast, Schurman et al. reported a graft half-life of 5.0 years for CAD reTx vs. 17.8 years for primary CAD (6). The reasons our results may differ from those of Schurman et al. may be related to center volume effect and patient mix (fewer non-Caucasians, different Tx era, more LDs) (19).

In summary, we report the results of the largest single-center study to date to assess variables predictive of long-term graft survival in children. We report substantial long-term graft survival rates in pre-adolescents. We show an association of primary disease recurrence with poor graft survival beyond 1 year postTx. We also show a substantial increase in long-term survival of grafts from LD compared to earlier reports. Only two of the independent variables affecting graft survival can be manipulated. Our data suggest that an emphasis on prevention of rejection, especially in the higher risk categories of CAD Tx and patients with diseases prone to recurrence, as well as preferential use of LD source, would have a major impact on long-term graft survival in children. Given the current advances in immunosuppressive therapy and postTx management, reTx may not be required for most of the productive lives of children without acute rejection.

Acknowledgments

This study was supported in part by NIH award AM13083. Dr Abhay Vats is supported by a National Kidney Foundation young investigator grant. We would like to thank Bobbi Wiercioch and Sandy Cragg for secretarial assistance and Jerry Vincent for illustrations.

Conflict of interest disclosure: Dr Abhay Vats is a consultant for ViraCor Biotechnologies.

Ancillary