The Honeymoon Phase and Studies of Nonsimultaneous Chains in Kidney‐Paired Donation

Kidney-paired donation programs have been successfullyincorporatingnondirecteddonorsintononsimultaneousex-tended (NEAD) chains (1). Including nondirected donors ineither NEAD chains or domino-paired donations (DPDs,chains that end in the waiting list rather than a bridgedonor) increases match opportunities for incompatiblepairs. When interpreting the performance of NEAD chainprograms,though,wemustrecognizethatthehoneymoonphase (where many transplants are facilitated by a singledonor)maynotlastforever.Aftertheinitialmonthsoryearsof operation, bridge donors will accumulate, competing forscarcematchesordroppingoutaltogetherafterlongwaits.Clinical experience in the longest-running registries alsoshows that many chains do not operate indefinitely, butdissipate as hard-to-match bridge donors languish.A recent report of Ashlagi et al. (2) highlights the need torecognize the honeymoon phase, as the data demonstratethat inferences from the early stages can be overturnedand reversed as registries operate over longer periods oftime. Ashlagi et al. reproduced our computational findings(3) that NEADs would not facilitate more transplants thanDPDs over 24 periods, or months, of operation. Then, us-ing 8-period simulations with longer segments (our sim-ulation limited concurrent segments to three pairs with-out limiting overall chain length), Ashlagi et al. argued thatNEADs with longer segments facilitate more transplantsthan DPDs. However, comparing Figure 2 with Figure 3 oftheir manuscript illustrates that the ostensibly minor detailofswitchingfroma24-periodsimulationtoan8-periodsim-ulationwasactuallythedecisivefactorinreversingthepre-dicted benefit of DPDs over NEAD chains, not the longersegments.Our simulations showed that after 24 periods in matureregistries, NEAD is not superior to DPD when segmentlengths are limited to three. Figure 2 of Ashlagi et al. re-produces this result, with the endpoint of the NEAD-3 linebelow 1 (near 0.98). Figure 3 shows the opposite resultfor the exact same experiment when simulating only thefirst eight periods (the honeymoon phase): the endpoint ofthe NEAD-3 line in Figure 3 is above 1 (near 1.015). Thatis, NEAD chains seem superior to DPDs in Figure 3 butnot in Figure 2, where the only difference is the number ofperiods simulated.Contrary to their honeymoon phase results, the data thatAshlagi et al. present actually prove that in mature reg-istries, the advantage of DPD over NEAD is larger withchains that allow longer concurrent segments. Figure 2compares DPD with NEAD for segments of length 4 andshows that for all renege rates, DPD gives an equal orgreater number of transplants than NEAD. Figures 3–10show valid data for honeymoon phase inferences, but can-not elucidate what will happen as NEAD chains deteriorateover time.Matching algorithms are also susceptible to the honey-moon phase problem. Ashlagi et al. limited their studies toonly 8-period simulations because, as they stated, optimiz-ing matches for later periods using longer segments wascomputationallyprohibitive.Thisraisesaquestion:howwilloptimizedmatcheswithlongersegmentsbecomputedforreal clinical registries beyond 8 periods? Of course, longersegments can also be logistically prohibitive because theconcurrent segments become harder to manage, so theadvantage of nonsimultanaeity (one of the strongest argu-mentssupporting NEAD chains) is lost. In practice, no con-current segments of a NEAD chain longer than three pairshavebeenreported.Consideringthesechallenges,limitingconcurrent segments to three pairs seems reasonable andmay even be necessary.We maintain that NEAD chains do not, in the long run,create more transplants than DPD because bridge donorsaccumulate in the system and compete against other po-tentialmatches.Inthelaterstagesofoursimulations,mostbridge donors did not find matches and some droppedout as their waiting time lengthened. We urge the trans-plant community to recognize the changing compositionof kidney-paired donation registries over time, beyond thehoneymoon phase, and prepare to meet the new chal-lenges that early success may bring.