Dear Editor:

Kidney transplant is the treatment of choice for patients with irreversible kidney failure, and the use of pretransplant flow cytometry crossmatch (FCXM) can improve outcomes in highly sensitized kidney transplant recipients.¹ In a FCXM, median channel fluorescence (MCF) shift or simply channel shift (CS) is calculated by subtracting the MCF of negative control serum from the MCF of recipient serum. In the absence of a standardized method to develop CS cutoffs to classify a T-cell-positive or B-cell-positive FCXM, the cutoffs are determined by individual laboratories by evaluating a negative control serum tested against lymphocytes from 30 healthy individuals or more, resulting in a derived mean and standard deviation (SD) of the MCF.^2-5 When data are normally distributed, 68.3%, 95.4%, and 99.7% of the distribution falls within 1SD, 2SD, and 3SD of the mean.6 Laboratories choose 2SD or 3SD as the CS cutoff.^2-4 When data are not normally distributed, highly skewed data can result in a high SD, potentially resulting in a false-negative FCXM.⁴

So far, there are no published reports on an alternate approach for generating a CS cutoff when data are not normally distributed. Here, we suggest the following 2 approaches: (1) median absolute deviation (MAD) and (2) interquartile range (IQR). For MAD, the median + 2MAD is calculated using 2MAD as a CS cutoff, which is more robust to outliers. In addition, MAD-based methods are expected to result in fewer false negatives.7 For IQR, IQR represents the range from the 25th percentile to the 75th percentile of a given data set, with IQR calculated by subtracting the 25th percentile from 75th percentile (Table 1).

As shown in Table 1, our laboratory generated 2SD, 2MAD, and IQR CS cutoffs using day-to-day FCXMs and American Society of Histocompatibility and Immunogenetics proficiency testing. Negative control serum used in the FCXM was from a normal human male donor with AB blood group and without anti-HLA class I and II immunoglobulin G antibodies (BioIVT, Westbury, NY, USA). As shown in Table 2, serum specimens from 3 renal transplant patients were tested against their donors’ T cells and B cells in a FCXM. Serum specimens were screened with the use of the LABScreen Single Antigen Assay (One Lambda, West Hills, CA, USA), which had a positive cutoff of ≥ 2000 mean fluorescence intensity (MFI). Using 2SD cutoff, we found that patient 1 serum with HLA class I and II donor-specific antibodies (DSA) and patient 2 serum with HLA class I DSA alone were incorrectly interpreted as T-negative/B-negative FCXM, whereas patient 3 serum with HLA class I DSA alone was incorrectly interpreted as T-negative/B-positive FCXM (Table 2). However, all 3 serum results were accurately interpreted as T-positive/B-positive FCXM using either 2MAD or IQR cutoff (Table 2).

It would be interesting to address in a prospective or retrospective study, where clinical outcomes, like antibody-mediated rejection, are correlated to any false-negative FCXM results with the use of 2SD, 2MAD, or IQR cutoffs. Because we report FCXM results to our transplant programs based on 2MAD cutoff, which is always lower than 2SD cutoff, every false-negative FCXM result from 2MAD cutoff is also a false-negative result from 2SD cutoff. Here, we compared the performance of each of the cutoffs using American Society of Histocompatibility and Immunogenetics proficiency testing, where the majority result or consensus result is taken as accurate. In the American Society of Histocompatibility and Immunogenetics proficiency testing comparison, serum with HLA class I DSA of 12 399 MFI against A32 resulted in B-cell FCXM with 75 CS. This would have put our laboratory in a minority result (24%) using 2SD cutoff of ≥ 78 CS but in a majority result (76%) using 2MAD or IQR cutoff of ≥ 42 CS (data not shown).

Overall, when the data are not normally distributed, 2MAD or IQR is superior to 2SD in preventing false-negative FCXM results, which have the potential to cause hyperacute rejection.

References:

1. Limaye S, O'Kelly P, Harmon G, et al. Improved graft survival in highly sensitized patients undergoing renal transplantation after the introduction of a clinically validated flow cytometry crossmatch. Transplantation. 2009;87(7):1052-1056.
   CrossRef - PubMed
   
2. de Moraes P, Fagundes I, Cardone JM, et al. Accuracy of the median channel shift in the flow cytometry for predicting complement dependent cytotoxicity crossmatching in kidney transplant candidates. Transpl Immunol. 2019;52(1):27-31.
   CrossRef - PubMed
   
3. Downing J. The lymphocyte crossmatch by flow cytometry for kidney transplantation. In: Christiansen F, Tait B, eds. Immunogenetics. Methods in Molecular Biology. Humana Press; 2012:379-390. Methods and Protocols; vol 882.
   CrossRef - PubMed
   
4. Dorwal P, Bansal SB, Chauhan R, et al. Median channel shift less than the cutoff in flow cytometric crossmatch: Not to be ignored! Asian J Transfus Sci. 2017;11(1):73-75.
   CrossRef - PubMed
   
5. Scornik JC. Detection of alloantibodies by flow cytometry: relevance to clinical transplantation. Cytometry. 1995;22(4):259-263.
   CrossRef - PubMed
   
6. Whitley E, Ball J. Statistics review 2: samples and populations. Crit Care. 2002;6(2):143-148.
   CrossRef - PubMed
   
7. Chung N, Zhang XD, Kreamer A, et al. Median absolute deviation to improve hit selection for genome-scale RNAi screens. J Biomol Screen. 2008;13(2):149-158.
   CrossRef - PubMed