Erythrocyte PIG‐A mutant frequencies in cancer patients receiving cisplatin

Abstract Background Cisplatin is a primary chemotherapy choice for various solid tumors. DNA damage caused by cisplatin results in apoptosis of tumor cells. Cisplatin‐induced DNA damage, however, may also result in mutations in normal cells and the initiation of secondary malignancies. In the current study, we have used the erythrocyte PIG‐A assay to evaluate mutagenesis in non‐tumor hematopoietic tissue of cancer patients receiving cisplatin chemotherapy. Methods Twenty‐one head and neck cancer patients undergoing treatment with cisplatin were monitored for the presence of PIG‐A mutant total erythrocytes and the young erythrocytes, reticulocytes (RETs), in peripheral blood for up to five and a half months from the initiation of the anti‐neoplastic chemotherapy. Results PIG‐A mutant frequency (MF) in RETs increased at least two‐fold in 15 patients at some point of the monitoring, while the frequency of total mutant RBCs increased at least two‐fold in 6 patients. A general trend for an increase in the frequency of mutant RETs and total mutant RBCs was observed in 19 and 18 patients, respectively. Only in one patient did both RET and total RBC PIG‐A MFs did not increase at any time‐point over the monitoring period. Conclusion Cisplatin chemotherapy induces moderate increases in the frequency of  PIG‐A mutant erythrocytes in head and neck cancer patients. Mutagenicity measured with the flow cytometric PIG‐A assay may serve as a tool for predicting adverse outcomes of genotoxic antineoplastic therapy.

][8][9][10] The sensitivity and throughput of the original rodent red blood cell (RBC) assay were improved by using fluorescent staining for detecting Pig-a mutant reticulocytes (RETs) and the development of methods for magnetic enrichment (ME) of mutants 11,12 or the RET fraction. 13As both the rodent and human assays are based on the same principle, similar sensitivity and throughput improvements were incorporated into the human assay. 14he performance of the rodent Pig-a assay was validated in interlaboratory trials and in multiple nonclinical safety studies [15][16][17][18][19] ; however, the performance of the advanced human PIG-A assay for detecting mutagenicity induced by exogenous agents has not been confirmed.Several studies that have been conducted using human RBC PIG-A assays were deficient in one way or another.It was demonstrated that workers operating in an environment where potentially hazardous chemicals (i.e., polycyclic aromatic hydrocarbons) are present have higher PIG-A MFs than workers operating in a clean environment, e.g., in the office. 20Other studies observed that cancer patients treated with anti-neoplastic chemotherapies on average have higher PIG-A MFs than selfidentified healthy people. 21,22But these studies did not measure PIG-A MFs in the same subjects before being exposed to hazardous environments or anti-cancer drugs, and causality between the exposure/treatment and elevated PIG-A MFs was not demonstrated directly.In the studies where PIG-A MFs were measured before and after exposure, the number of study subjects was insufficient for making conclusions. 23n the current study, we have verified the performance of the advanced human erythrocyte PIG-A assay in a longitudinally controlled study.RET and total RBC PIG-A MFs were evaluated in a large group of cancer patients before the start of chemotherapy, at several time points during chemotherapy, and after the completion of the chemotherapy regimen.
The study had several objectives: 1.To determine if the human erythrocyte PIG-A assay is sufficiently sensitive for detecting mutagenicity caused by therapeutic doses of a known genotoxin, cisplatin, which has been shown to induce Pig-a mutant erythrocytes in the rodent assay; 2. To determine if the success or failure of cisplatin anti-cancer effect is related in any way to the change in erythrocyte PIG-A MF during the chemotherapy regimen; and 3. To lay a foundation for a long-term study associating the value of PIG-A MF in response to the relatively short-term treatment with cisplatin and the potential for tumor recurrence or induction of secondary tumors in the same cancer patients in the future.

| Study subjects
All procedures related to patient recruitment for the study, procedures for handling biological samples, and the patient informed consent form were reviewed and approved by the University of Arkansas for Medical Sciences (UAMS) Institutional Review Board and the FDA Research Involving Human Subjects Committee.Patient recruitment occurred at the UAMS Cancer Institute.The recruited patients were allowed to leave the study at any stage without any precondition.Study subjects were newly diagnosed cancer patients, for whom cisplatin chemotherapy was prescribed as a part of the complete therapy regimen.Patients receiving concomitant targeted radiotherapy were included in the study population as long as major sites of hematopoiesis were not exposed to ionizing radiation.Head and neck cancer (HNC) patients recommended for a standard 3 × 100 mg/m 2 cisplatin infusion chemotherapy regimen, with or without radiation, were the preferred cohort.The study subjects agreed to provide a blood sample at each cisplatin infusion cycle and an additional sample at a follow-up visit 1-5 months after the last infusion cycle (for a total of four blood samples).Peripheral blood was collected at the Cancer Institute; approximately 6 mL of peripheral blood was drawn from the saphenous vein into a standard EDTA-K 2 Vacutainer tube.

| Erythrocyte PIG-A analyses
Phosphatidylinositol glycan class-A analyses were performed at the National Center for Toxicological Research (NCTR).In most cases, blood was processed within 2-3 h of collection.In cases when the blood sample was collected late in the day, the vacutainer tube with blood was stored overnight at +4°C and processed for the PIG-A assay the next morning.Erythrocyte PIG-A assays were performed using the In Vivo MutaFlow PLUS-25H kit (Litron, Rochester, NY) following the manufacturer-suggested protocol.Gradient centrifugation on Lympholyte®-H cell separation medium (Cedarlane, Burlington, NC) was used for leucodepletion.Anti-PE magnetic beads, LS magnetic columns, and quadroMACS™ magnet were used for magnetic enrichment of the mutant fraction (all from Miltenyi Biotec, Gaithersburg, MD).CountBright™ beads (Thermo Fisher Scientific; Waltham MA) were used in flow cytometric analyses for determining cellequivalents and calculating RET and total RBC MFs as suggested by Litron.Flow cytometry was performed on a FACSCanto™ II analyzer equipped with 488nm and 633-nm lasers and operating under the control of FACSDiva™ v.9 data acquisition software (BD Biosciences; San Jose, CA).All PIG-A assays were performed in triplicate to increase the accuracy of MF measurement.Each replicate utilized 200 μL of anticoagulant-preserved blood (for a total of 600 μL of blood for the three replicates assayed at each time point).In a typical blood sample, each replicate generated mutant frequency data on 2-3 × 10 8 total RBC equivalents and 3-8 × 10 6 RET equivalents.

| Statistical analyses
The average values of observed triplicate values of the RET MFs and total RBC MFs for each time point were used in analyses and log-transformations.MFs were accessed for normality violations using the Kolmogorov-Smirnov, Cramer-von Misses, and Anderson-Darling tests.A one-way analysis of variance (ANOVA) mixed model was used to evaluate the relationship between PIG-A MF and observation time.Pairwise multiple comparisons between the observation time points were evaluated using the post hoc Dunn-Šidák correction to control the Type I error rate.The non-parametric Jonckheere-Terpstra test was used to evaluate the overall trend in PIG-A MFs over increasing time post-exposure to cisplatin.All statistical tests were two-sided, unless otherwise stated, and significance was assessed at the nominal 0.05 level.SigmaPlot 14.5 (Inpixon; Palo Alto, CA) and SAS for Windows v9.3 (SAS; Cary, NC) software were used for statistical analyses.
To perform pairwise comparisons, PIG-A assay results were grouped in four categories based on the patients' office visit for blood donation and/or cisplatin infusion.Group 1 (pre-treatment; N = 21) were blood samples from patients visiting the oncology clinic for the first cisplatin infusion and the first blood collection (Day 0 of the study; cisplatin infusion was administered after blood collection).Group 2 (N = 21) were samples from the same patients during their visit for the second cisplatin infusion (approximately Day 21 of the study; the second blood sample was collected before cisplatin was administered).Group 3 (N = 20) samples were from the third office visit of patients that remained in the study (approximately Day 42 of the study; cisplatin infusion was administered on this day but not to all patients).Group 4 (N = 15) were from the fourth visit of patients that remained on the study (Days 70-159 of the study; no infusion was administered on this visit).Grouping was proportional but not necessarily linear with regards to cumulative dose of cisplatin received (see Table 1 for details of cisplatin treatment).

| RESULTS
Twenty-four cancer patients were recruited for the erythrocyte PIG-A study (see Table 1 for complete description).Twenty-three patients were diagnosed with various stages of HNC, and one female was diagnosed with cervical cancer.All cases (including the cervical cancer) were typed as squamous cell carcinoma.The patients were from different ethnic backgrounds (16 white, 6 black, and 2 other) and ranged from 32 to 86 years of age.Eighteen patients were males and six were females; 16 patients were identified as former or current smokers.For all HNC patients, a standard chemotherapy regimen consisted of 3 cycles of 100 mg/m 2 cisplatin infusion (one infusion every 3 weeks).The patient with cervix squamous cell carcinoma received a different chemotherapy regimen, with more frequent infusions of lower cisplatin doses.Three patients withdrew from the study after the first cycle of cisplatin infusion, submitting only a Day 0 blood sample; these three patients were not included in the analyses.Not all HNC patients received the complete prescribed regimen of cisplatin for various reasons (some withdrew from the study, while others refused the final dosing).Only 15 patients submitted the requested four blood samples (including the sample at a follow-up visit).Only 11 patients received all three prescribed cycles of cisplatin infusion (at regular intervals) and provided the requested 4 blood samples.The schedule of blood collections relative to cisplatin infusions for all patients is shown in Figure 1.Of the 21 patients remaining on the study after the first cycle of cisplatin infusion, 15 did not show evidence of disease at the end of the study; some of these did not complete the full 3-cycle chemotherapy regimen.In three patients, the response to cisplatin therapy was either partial or negative; in another three patients, the outcome of chemotherapy could not be determined because they withdrew from the study after the second or third infusion cycle and didn't return for the follow-up visit.
Reticulocyte and total RBC PIG-A MFs determined at time points the patients consented for submitting blood samples are shown in Table 2.The raw RET and total RBC PIG-A MFs for patients grouped by office visit are shown in Figure 2. Also, fold-change values relative to the pretreatment baseline values were computed for exploratory visual assessments (see Figure 3).No statistical tests were performed on this measure; however, a fold-change value of about 2 was considered of interest.In most patients, PIG-A MF was increased in both fractions of RBCs at one or more time points relative to the pre-treatment time point, Day 0. A somewhat earlier response in the induction of PIG-A MFs was observed for the RET fraction.
Descriptive summary statistics are shown in Table 3. Raw MF data were not supportive of bell-shaped normal distribution, as all applied normally tests were significant (p < 0.05).Hence, the MF data were log-transformed to conform with the normal distribution.The mixed model ANOVA was the preferred method of analysis because of its robustness to slight deviations from normality and its flexibility to account for missing values under the assumption of "ignorable missingness", i.e., missing at random or missing completely at random. 24Generally, there were positive linear trends in RET PIG-A MF (p < 0.001) and in total RBC PIG-A MF (p = 0.007) over time following cisplatin infusions (Table 3).The greatest increases in MFs for the RET fraction were noted at visit 3 (at Day 42 or more) and visit 4 (at Day 70 or more), and at visit 4 for the total RBC fraction.
The overall effect of cisplatin infusion exposure over time was significant for RET PIG-A MF (p < 0.001) and total RBC PIG-A MF (p < 0.0037) as determined by ANOVA mixed model analysis.The results of pairwise multiple comparisons between mean mutant frequencies at each time-point following cisplatin infusions are summarized in Table 4.The RET MFs were significantly elevated in cancer patients at three or more weeks after the initial infusion of cisplatin (comparisons of Day 0 vs. Day 21, p = 0.0006; vs. Day 42, p = 0.0021; and vs. Day 70, p < 0.0001).The effect of cisplatin treatment on the total RBC PIG-A MFs was not evident after the first and the second infusions (comparisons of Day 0 vs. Day 21, p = 0.1550; and vs. Day 42, p = 0.1041).Total RBC PIG-A MFs became Note: In all patients the type of cancer was identified as squamous cell carcinoma.Ethnic origin is believed to be irrelevant for this study; this information is not shown in the Table .Patients SL006, SR007, and TJ008 withdrew from the study after the first cycle of cisplatin therapy.
significantly increased relative to prior time-points only at follow-up visits at Day 70 or later of the study (comparisons of Day 70 vs.Day 0, p = 0.0032; vs. Day 21, p = 0.0106; and vs. Day 42, p = 0.0081).These observations are consistent with the presumption that in the human erythrocyte PIG-A assay, RETs are the leading fraction for detecting mutagenicity and total RBCs are the lagging fraction, as is the case for the rodent Pig-a assay.

| DISCUSSION
Cisplatin, cis-diammine dichloroplatinum (II) or cis-PtCl 2 (NH 3 ) 2 , is a primary chemotherapy choice for various solid tumors, including HNC, one of the top 10 types of cancer worldwide.The most common cisplatincontaining treatment regimen for HNC, as recommended by the Radiation Therapy Oncology Group, is 100 mg/m 2 on Days 1, 22, and 43, in combination with radiotherapy (reviewed in Brown et al. 25 ).In many cases (and as seen in the current study), initial cisplatin chemotherapy produces immediate positive results, i.e., no evidence of disease at the completion of the therapy course.
In an aqueous environment, chlorine molecules can be displaced from cisplatin by water to produce positively charged mono-or di-substituted complexes that serve as electrophiles for attack of electron-rich N7 purine positions of genomic DNA.The most frequent tin may be a biomarker for the future recurrence of cancer or the emergence of secondary malignancies.An increase in MF may also be a biomarker for the responsiveness or resistance of secondary or recurrent tumors to future cisplatin chemotherapy.The erythrocyte PIG-A assay is one of the few practical assays for assessing mutagenesis in human cancer patients.Very little blood is required for performing the assay.As cisplatin is administered to patients as an intravenous infusion, it is certain that cisplatin reaches the bone marrow, where it can induce mutations in nucleated dividing erythroid precursors cells.Previous studies indicate that cisplatin induces mutant cells in the rat erythrocyte Pig-a assay, [27][28][29] thus, it is likely that cisplatin is also a human PIG-A gene mutagen.However, the optimal timing for measuring erythrocyte PIG-A MFs after exposure to a potential mutagen is not known, making it necessary to measure PIG-A MFs at several time points.In the current study, these time points ranged from as early as possible after the first infusion of cisplatin to after completing the chemotherapy regimen.
Our human PIG-A study did not contain a separate vehicle-control group of subjects.For each patient, the sample processed at Day 0 before the first cisplatin infusion served as control for that patient.From the literature, it is known that background PIG-A MFs in individual healthy human subjects remain within a narrow range when measured repeatedly over a period of months; however, the background MFs measured in different human subjects vary widely.Self-identified healthy donors have total RBC PIG-A MFs from 1 × 10 −6 to 10 × 10 −6 , with occasional outliers having MFs from 100 × 10 −6 to F I G U R E 1 Blood drawing and cisplatin infusion timing relative to day of the study.Not all subjects submitted the requested four blood samples; not all subjects completed cisplatin chemotherapy course.The therapy course was not the same for all subjects.The vertical order of patients is based on the number of donated blood samples and the number of cisplatin infusions.Groups were based on office visits for blood sampling.The 11 patients highlighted in bold received the standard 3-infusion cisplatin therapy course and submitted the requested four blood samples.For details, see Results and Table 1.
T A B L E 2 Frequencies of PIG-A mutant reticulocytes (RETs) and mutant total red blood cells (RBCs) in peripheral blood of cancer patients receiving cisplatin chemotherapy.

Patient ID and response to cisplatin b
Day of blood draw RET MF a (×10 Note: The total of 79 PIG-A assays were performed in triplicates; only one replicate failed (CM019, Day 42).The first three blood collections occurred on days of cisplatin infusions (i.e., on Days 0, ca.Day 21, and ca.Day 42), unless the third infusion was canceled for whatever reason.When blood collection day coincided with cisplatin infusion day, blood was drawn before the infusion.Blood collections post-Day 43 occurred at follow up visits; no infusions were performed on these days.Not all patients provided the expected four blood sample and not all patients received the prescribed three infusions of cisplatin.KD009 had a different schedule of cisplatin infusion, and the third blood sample was collected on Day 36 instead of ca.Day 42.LH017 had 4 cycles of cisplatin infusion; the blood was not collected at the fourth infusion cycle.Maximum increase in RET MF relative to MF determined on Day 0 was recorded for WM003 on Day 43 of the study (~20 × 10 −6 ); maximum increase in total RBC MF was recorded for CC016 on Day 155 (~11 × 10 −6 ).Maximum fold-increases in MFs relative to Day 0 were recorded for different patients (see Figure 3).a Mutant frequency, MF. b Response to cisplatin chemotherapy is clarified in Table 1.
c No cisplatin infusion happened during the third visit; see explanation in Table 1.
Another complication for the study was that not all patients were able to provide the complete set of four blood samples that were requested at the start of the study.Thus, the original study protocol of 3 standard infusions of 100 mg/m 2 cisplatin combined with 4 blood collections at predetermined time points (0, 3 weeks, 6 weeks, and 10 to 14 weeks) was not followed for many of the study subjects.There are various approaches for the statistical analyses of such incomplete datasets (e.g., paired t-test or repeated measures ANOVA), but they would require performing analyses on reduced sub-sets of patients, which would inevitably decrease the power of analyses.Hence, the ANOVA mixed model was employed to use the full PIG-A MF datasets that were acquired.Despite all the caveats and logistical difficulties in performing the study, we were able to detect a general trend for increasing PIG-A MFs in cisplatin-treated patients (see Table 3).However, the response to treatment in individual patients was relatively small: the maximum relative increase in MF in the RET fraction was under 6-fold, and the maximum absolute increase was approximately 20 × 10 −6 .The maximum increase in MF for the total RBC fraction was less than 5-fold, and the maximum absolute increase was just under 11 × 10 −6 .In one patient, DW012, neither RET nor total RBC PIG-A MF increased over the study period.Similar modest increases in PIG-A MF were observed in a small group of testicular cancer patients receiving cisplatin-containing chemotherapy. 30However, the chemotherapy regimen in the testicular cancer study was different than in our HNC study.
The response in the PIG-A assay to cisplatin treatment in cancer patients may reflect specific properties of the human erythrocyte PIG-A assay.The PIG-A assay was developed as an extension of a diagnostic procedure for characterizing PNH.In PNH, a mutation in the endogenous X-linked PIG-A gene in hematopoietic bone marrow stem cells causes a disruption in the synthesis of glycosyl phosphatidyl inositol (GPI), which makes erythrocytes originating from the mutant bone marrow stem cells deficient in CD55 and CD59 GPI-anchored surface proteins vulnerable for complement-mediated lysis. 31As the CD59/CD55-deficient phenotype is measured in the human erythrocyte PIG-A assay, it might be hypothesized that cisplatin-induced GPI-deficient PIG-A mutant erythrocytes in cancer patients are under pressure for elimination by the complement system, as occurs in PNH patients.
The dynamics of PIG-A mutant erythrocytes accumulation (due to mutagenicity with cisplatin or any other xenobiotic agent) and elimination of PIG-A erythrocytes (due to immune surveillance) are not known for humans.In the case of PNH, evidence indicates that even the youngest cohort of PIG-A mutant erythrocytes, RETs, is under pressure for complement-mediated elimination. 32However, several lines of evidence suggest that such elimination of mutants is neither rapid nor 100% efficient: PNH patients have high levels of CD55/CD59-deficient RBCs in circulation. 33Also, several cases have been documented of relatively high frequencies of PIG-A mutant RBCs in self-identified healthy donors, as well as in some cancer patients.In the current study, we used a version of the PIG-A assay that detects mutants among total RBCs and RETs.The assumption was that PIG-A mutant RETs are exposed to the complement system for a shorter period of time than are mature RBCs, so that RETs may be more sensitive than total RBCs for F I G U R E 2 Box and whiskers percentile charts for reticulocyte (RET) and total RBC absolute PIG-A mutant frequencies (MF) measured in cancer patients receiving cisplatin.The boundaries of the box represent the 25th and 75th percentiles; the horizontal line within the box marks the median; and the whiskers indicate the 10th and 90th percentiles.MFs were determined at scheduled office visits for cisplatin infusions or for follow-up post-therapy visits at which blood samples were collected (Groups 1, 2, 3, and 4; see Table 1 and Materials & Methods for details on timing of office visits).MFs are shown for all 21 patients (fewer blood samples were collected at the third and the fourth office visits as some patients withdrew from the study; see Table 1).Data for the charts are taken from Table 2.
detecting the mutagenic effects of cisplatin chemotherapy.Our observations, using an advanced PIG-A assay employing magnetic enrichment of mutant erythrocytes, suggest that the mutagenic effects of cisplatin chemotherapy in human patients can be detected in RETs as early as 3 weeks after the first exposure, while in total RBCs, cisplatininduced increases in MF were detected only at 10 weeks or longer after the first exposure.
Of interest is that, in the rat cisplatin study, Pig-a MFs increased approximately 5-fold in high-dose treated animals, and the absolute increases in MF (approximately 20 × 10 −6 ) were similar for all animals in the treated group. 27Perhaps the more robust and uniform Pig-a response observed in the rat model of cisplatin-induced mutagenesis was at least partially due to lack of complement-mediated hemolysis of Pig-a mutant RBCs.The maximum cumulative dose of cisplatin in the rat study (6 mg/kg) was comparable to the total dose used in this human study (3 × 100 mg/m 2 ), but the rats did not have tumors, and a more acute regimen of cisplatin exposure was used.Yet, a non-linear increase in RET MF was observed for the rats treated with a low dose of cisplatin, 27 just as was observed for several subjects in our study (e.g., RJ002, AW010, MM011).
In conclusion, we observed moderate increases in erythrocyte PIG-A MF in patients receiving cisplatin chemotherapy.This confirms that the human and rodent Pig-a/Pig-a T A B L E 3 Descriptive summary statistics and trend analysis for RET and total RBC PIG-A MFs.assays measure the same relevant endpoint (mutation), while the rodent assay may be a bit more sensitive due to better control over the experimental conditions and timing for blood analyses.In our study, there was no clear relationship between cisplatin-induced PIG-A MFs and the shortterm outcome of chemotherapy; most patients exhibited no evidence of disease at the end of the study.Long-term monitoring of the study subjects, including measuring PIG-A MFs, may shed a light on the role of mutagenesis and DNA repair in tumor recurrence, the induction of secondary tumors, and resistance to cisplatin chemotherapy.

Z c = 3 .
997, N d = 77, p-value e < 0.001 Z c = 2.452, N d = 77, p-value e = 0.007 a Standard error of mean.b Mean of fold change.c Value for Z-statistic.d Sample size for the estimated Z-statistic.e One-sided p-value.

F
I G U R E 3 Time-dependent relative fold-increases in RET and total RBC PIG-A MFs in cancer patients receiving cisplatin chemotherapy.Each data point was calculated as the ratio of the PIG-A MF determined at a specific day of study and the MF determined at Day 0 (the first office visit) before cisplatin infusion (data for the charts are taken from Table2).Only the numerical index of the patient ID and the disease status at the end of the study are shown in the charts.Data points for each patient are traced by lines.Dotted lines represent patients with unknown status of disease at the end of the study.
Patient demographics, diagnostics, and treatment with cisplatin chemotherapy.
T A B L E 1

and response to cisplatin b Day of blood draw RET MF a (×10 −6 ) Total RBC MF a (×10 −6 )
12,23 10 −6 .12,23Therefore,groupinghumansinto control and treated groups could result in poor resolution between background PIG-A MFs and cisplatin-induced PIG-A MFs.In addition, cancer patients in the current study varied by age, and finding age-matched control subjects would be burdensome for the study.Patient ID

contrasts RET PIG-A MF Total RBC PIG-A MF F [3,20] a = 13.46; p-value b < 0.0001 F [3,20] a = 6.21; p-value b = 0.0037 t-value c DF d Adjusted. p e t-value c DF d Adjusted p e
Statistical comparisons of RET and total RBC PIG-A MFs in patients receiving cisplatin chemotherapy.Values adjusted for multiple comparisons using the Dunn-Šidák correction.
T A B L E 4 a Estimated value of the F-test statistic [numerator DF = 3; denominator DF = 20].b p-value for one-way ANOVA mixed model analysis.c Estimated value of the t-test statistic.d Degrees of freedom.e p-