Impact of prior melphalan exposure on stem cell collection in light chain amyloidosis

Abstract
Use of melphalan in multiple myeloma was observed to have a deleterious effect on stem cell collection in older studies. There is limited data on the impact of melphalan in light chain (AL) amyloidosis, especially in the plerixafor era. We retrospectively evaluated stem cell mobilization in 610 patients with AL amyloidosis, of which 79 had prior exposure to melphalan, 167 to other chemotherapeutics, while 364 had no chemotherapy exposure. Collection of CD34+ stem cells × 106/kg was lower in the melphalan group. Median total yields in the melphalan, non-melphalan, and no chemotherapy groups were 5.5, 7.7, and 7.8, respectively; p < 0.001. Day-1 yields were 2.7, 3.5, and 4.0 (p = 0.0003), respectively, and median yields per collection were 2.0, 3.3, and 4.0 (p < 0.001), respectively. Similar results were observed in the sub-group analysis after plerixafor was integrated in our collection algorithm (2009). Patients in the melphalan group had higher failure rate of 9% vs. 2% each in the other two groups (p = 0.006). Impact of melphalan was dose-dependent, with cumulative melphalan exposure of >150 mg (median: three cycles) resulting in lower yields. Therefore, duration of melphalan exposure prior to stem cell collection should be limited, ideally, not exceeding more than two cycles of treatment.

Introduction
High-dose melphalan and autologous stem cell transplant (ASCT) with appropriate patient selection is an integral part of treatment of light chain (AL) amyloidosis [1, 2]. Prior to proceeding with transplant, patients may receive a few cycles of chemotherapy, often with melphalan or bortezomib-based regimens. This is usually done for many reasons including timely initiation of plasma cell directed therapy if delays are anticipated prior to transplant, bone marrow debulking for patients with high plasma cell burden, or if a patient is a borderline candidate for trans- plant at diagnosis and becomes eligible later [1]. Currently, it is unclear whether there is any difference in outcomes based on melphalan or bortezomib-based initial therapy, and therefore the choice of chemotherapy is often guided by institutional guidelines and patient and physician pre- ference. Even when bortezomib is added to upfront therapy, it may be combined with melphalan, although cyclopho- sphamide is used more often in the United States [3].
Previous studies have shown that melphalan adversely impacts CD34+ stem cell yield in patients with multiple myeloma in a dose-dependent manner [4–9], and therefore there is concern in the use of melphalan in transplant- eligible patients in AL amyloidosis. However, a randomized controlled trial in patients with multiple myeloma showed that an adequate number of CD34+ stem cells can be col- lected following a short course of alkylating agents (mel- phalan and cyclophosphamide) [10]. These studies were conducted many years ago and there have been significant advances in the process of stem cell collection in the meanwhile, most notable being the availability of plerix- afor. Plerixafor is a reversible antagonist of CXCR4. It inhibits binding of SDF-1α (stromal-derived factor-1α) to the CXCR4 receptor, which is essential for homing of CD34+ stem cells in the bone marrow microenvironment, thereby leading to mobilization of stem cells to the per- ipheral blood [11–13]. Studies have shown that the addition of plerixafor results in superior stem cell yields in AL amyloidosis [14, 15]. There is a lack of data on the impact of melphalan on stem cell yields in patients with AL amyloidosis in the current era. In this study, we have retrospectively evaluated the impact of melphalan on stem cell yield and engraftment kinetics in patients in AL amyloidosis relative to other chemotherapy regimens and in patients who had no expo- sure to prior chemotherapy.

Following Institutional Review Board approval, all patients with AL amyloidosis seen at our institution who underwent stem cell mobilization from January 2001 to March 2016 were identified. Only patients who had provided consent for review of their records for research purposes were included. Data regarding diagnosis, treatment, stem cell mobilization, and ASCT were collected from the electronic medical records.The standard goal of stem cell collection at our institu- tion is 3–4 × 106 CD34+ cells/kg per transplant, with a minimum of 2 × 106 CD34+ cells/kg. The mobilization regimen has evolved over the study period as described in Supplementary Figure 1. Prior to 2009, granulocyte colony-stimulating factor (G-CSF) at a dose of 10 mcg/kg was used for at least 4 days followed by peripheral blood CD34 count on day 5. If the CD34 count in peripheral blood was inadequate, higher doses of G-CSF (16 mcg/kg two times per day) or in some cases granulocyte macrophage colony- stimulating factor (GM-CSF) could be administered. Since 2009, we have implemented a risk-adapted mobilization algorithm, which includes G-CSF at 10 mcg/kg for at least 4 days with on-demand plerixafor [16]. High-dose cyclo- phosphamide is used rarely on a case-by-case basis, typi- cally for patients with high plasma cell burden.Total, day-1, and average per collection yields (total cells collected/number of collections) of CD34+ stem cells (×106) per kg are reported in the results section. Stem cell mobilization was deemed a failure if the patient did not collect adequate CD34+ stem cells to meet the collection goal or did not have adequate peripheral blood CD34 cells and the collection attempt was abandoned without apher- esis. Neutrophil engraftment was defined as the first of three consecutive days of absolute neutrophil count (ANC) of >500 × 109/L. Platelet engraftment was defined as the first of three consecutive days with platelet count of 20,000 × 109/L, without a platelet transfusion in the preceding 7 days. Time to platelet count of 50,000 × 109/L for three con- secutive days was also evaluated. Analysis was carried out using JMP® 12 (SAS Institute Inc., Cary, NC) statistical software [17]. Chi-Square and Fischer Exact tests were used to carry out univariate ana- lysis for categorical variables and Wilcoxon Rank Sum/ Kruskal Wallis for continuous variables. p-Values for two- sided α = 0.05 are reported. Patients were analyzed in three groups based on prior exposure to melphalan, non- melphalan chemotherapy, or no prior chemotherapy expo- sure. p-Value from comparison between all three groups is designated as p1 and the p-value from comparison between melphalan and non-melphalan groups is designated as p2.

Results
From January 2001 to March 2016, 610 patients with AL amyloidosis underwent stem cell mobilization. Of these, 364 (60%) underwent stem cell collection prior to receiving any chemotherapy or received steroids alone, 79 (13%) had prior melphalan exposure, and 167 (27%) were exposed to other agents. Baseline characteristics and mobilization regimens are described in Table 1. There was no difference in age and gender distribution amongst the three groups. In the melphalan group, patients were exposed to chemother- apy for a shorter period prior to stem cell collection relative to the non-melphalan group (median: 2 vs. 4 months, p<0.001). Table 2 describes the chemotherapy regimensreceived by patients prior to stem cell collection in the non- melphalan group. Overall, 51% (85/167) of patients were exposed to cyclophosphamide and 16% (27/167) to lenali- domide. Since 2009, plerixafor has been incorporated in our mobilization algorithm. A higher proportion of patients in the melphalan (39%) and non-melphalan (44%) che- motherapy groups required plerixafor in the 2009-onwardstime period compared to patients not exposed to any prior chemotherapy (20%), p1 <0.001. However, there was nodifference in plerixafor use amongst the two groups exposed to chemotherapeutic agents (p2 = 0.5).Table 3 and Fig. 1 describe the stem cell yields (CD34+ cells × 106/kg) amongst the three groups. Melphalan exposure was associated with significantly lower day-1 stem cell yield, average yield per collection, as well as, total stem cells collected compared to non-melphalan-based che- motherapy and no chemotherapy exposure. Median day-1 collection yields in the three groups were 2.7, 3.5, and 4.0, respectively (p1 = 0.0003 and p2 = 0.0007).

Median stem cells per collection were 2.0, 3.3, and 4.0, respectively (p1and p2 <0.001). Median total stem cells collected in the three groups were 5.5, 7.7, and 7.8, respectively (p1and p2<0.001). Across all groups, the median number of apheresis sessions required to meet stem cell collection goal was two. However, the range of apheresis sessions required to meet the collection goal was higher in the melphalan group compared to the other two groups, with interquartile (IQR) ranges in the three groups being 2–4, 1–3, and 1–3, respectively (p1 = 0.007 and p2 = 0.14). The rate of suc- cessful mobilization was significantly lower in the mel- phalan group (91%), compared to non-melphalan (98%) and no prior chemotherapy (98%) groups, p1 = 0.006 and p2 =0.04. Overall, 12 patients who failed the first mobilization attempt underwent a second attempt at stem cell mobiliza- tion. The second mobilization attempt was successful in two of five patients in the melphalan group, two of two patients in the non-melphalan chemotherapy group, and five of five in the no prior chemotherapy exposure group. Two patients in the melphalan group underwent a third mobilization attempt, which was successful in both cases.A sub-group analysis of patients who underwent stem cell collection after 2009 was carried out, as a risk-adapted pler- ixafor algorithm was implemented at that time point. Similar results were observed and patients who were exposed to melphalan prior to collection had lower stem cell yields, including day-1 collection, average yield per collection, and total stem cell collection as described in Supplementary Table 1. Median day-1 collection yields in the melphalan, non- melphalan, and no chemotherapy group were 2.5, 3.7, and 4.9, respectively (p1 <0.001 and p2 = 0.02). Median stem cells collected per day of apheresis in the three groups were 2.4, 3.4, and 4.8, respectively (p1 <0.001 and p2 = 0.007). Median total CD34+ yields were 5.3, 7.8, and 7.9, respectively (p1 and p2= 0.0002). The median number of apheresis sessions was twoacross all groups and stem cell collection was successful in 37/ 39 (95%) patients exposed to melphalan, in 134/138 (97%) patients exposed to other chemotherapeutic regimens, and in 158/159 (99%) patients who collected stem cells without any exposure to chemotherapy (p1 = 0.1 and p2 = 0.5).

In all patients with prior exposure to melphalan, the median number of melphalan chemotherapy cycles prior to stem cell collection was two (range 1–20). The route of melphalan administration was oral in 63 patients (80%), intravenous in 15 patients (19%), and was not known in one patient. Exact or weight-based dosing per cycle was known in 69/79 (87%) patients. In the remaining 10 patients, dose was estimated using standard dosing for melphalan used for the regimen received and multiplying it by the number of chemotherapy cycles. Median estimated cumulative dose of melphalan received prior to stem cell collection was 140 mg (range 22–1680). To determine if there was a dose–effect relationship with melphalan exposure and stem cell yield, we evaluated stem cell yields in patients with cumulative melphalan exposure of more than 150 mg compared to lower cumulative doses, as described in Table 4. Patients exposed to more than 150 mg of melphalan had lower day-1 CD34+ stem cell yield (2.2 vs. 3.0, p = 0.006), lower stem cell yield per collection (1.4 vs. 2.5, p = 0.002), and lower total stem cell yield (5.0 vs. 5.7, p = 0.03). The mediannumber of apheresis sessions were three and two, respec- tively, p = 0.04.Stem cell collection yields in patients exposed to cyclo- phosphamide and lenalidomide are described in Table 5. Patients in the cyclophosphamide group had higher yields compared to melphalan-exposed patients, including day-1 yield (3.6 vs. 2.7, p = 0.006), average yield per collection (3.5 vs. 2.0, p = 0.006), and total yield (7.1 vs. 5.5, p = 0.002). A trend toward less apheresis sessions was also observed (median of two sessions in both; IQR of 1–3 vs. 2–4, p = 0.06). Mobilization was successful in 100% vs. 91% patients, respectively (p = 0.005). There was no dif- ference in stem cell yields when patients exposed to cyclophosphamide were compared to those without any prior chemotherapy exposure (Supplementary Table 2). Twenty-seven patients had prior exposure to lenalidomide. Total stem cell yields (8.5 vs. 7.8, p = 0.4) in this group were similar to patients who were not exposed to any chemotherapy. However, there was a trend toward lower day-1 yield (3.0 vs. 4.0, p = 0.08) and average yield per collection (3.1 vs. 4.0, p = 0.1) in the lenalidomide group.

Median number of apheresis sessions were three (IQR 2–4) vs. two (IQR 1–3) (p = 0.007) and mobilization was suc- cessful in 93 vs. 98% patients, respectively (p = 0.1) (Supplementary Table 3).Overall, 508 of the 610 patients (84%) who attempted stem cell collection underwent a transplant at some point in their disease course. Of these, 26 had collected stem cells after exposure to melphalan, 140 after exposure to other chemotherapeutic agents, and 342 did not have any prior exposure to chemotherapy. The proportion of patients undergoing stem cell transplant was lower in the cohortpreviously exposed to melphalan (26/79; 33%) compared to other chemotherapy regimens (140/167, 84%), p <0.001. Details of stem cell transplant and engraftment kinetics are described in Table 6. The median time from collection to day 0 was longer in the melphalan group compared to the non-melphalan and no chemotherapy groups (31 vs. 9 vs. 9 days, p1 and p2 <0.001; excluding first collection failures: 15 vs. 9 vs. 9 days, p1 <0.001 and p2 = 0.0001). Total CD34+ stem cells infused were lower in the melphalan group (3.6 vs. 4.2 vs. 4.4, p1 <0.001 and p2 = 0.04) com- pared to the other two groups. There was some indication of longer time to engraftment in the melphalan group, though absolute number of excess days was only 1–2 compared to the other groups. Time to platelet engraftment in the mel- phalan, non-melphalan, and no chemotherapy groups was 19 vs. 17 vs. 18 days, respectively; p1 and p2 = 0.03. Time to neutrophil engraftment was 15 vs. 16 vs. 14 days, respectively; p1 = 0.02 and p2 = 0.4. Overall, median time to both neutrophil and platelet engraftment was 19 days in the melphalan group, 17 days in the non-melphalan che- motherapy group, and 18 days in the no chemotherapy group, p1 = 0.08 and p2 = 0.04.

Discussion
Our study demonstrates that prior exposure to melphalan impairs bone marrow reserve and results in lower stem cell yields in patients with AL amyloidosis, and there appears to be a dose–effect relationship. This holds true even with the availability of plerixafor, which has led to superior stem cell collection outcomes across various hematologic malig- nancies [12–15].We observed that patients who are exposed to melphalan have lower yields of stem cells, including total stem cells collected, day-1 yield, and average yield per apheresis session as compared to non-melphalan based chemotherapy exposure and no chemotherapy exposure. This is despite the fact that median duration of exposure to melphalan before stem cell collection was 2 months compared to 4 months forother regimens (p <0.001). Shorter duration of chemother-apy exposure in the melphalan group likely reflects our practice pattern based on data from prior studies in multiple myeloma demonstrating the deleterious effect of melphalan on stem cell collection. It is important to note that mel- phalan exposure resulted not only in lower stem cell yields, but also a higher failure rate of 9% in the melphalan group, compared to 2% seen in the other two groups. Even with the availability of plerifaxor, failure rate in the melphalan group was 5%. There was also an indication of longer time to achieve engraftment, though the absolute difference in number of days was only 1–2 and is likely not clinically relevant. It is interesting to note that the proportion of patients who ultimately underwent stem cell transplant was lower is the melphalan group compared to the non-melphalan group (33% vs. 84%, p <0.001).

It is possiblethat the treating physicians may have been more proactive about collecting stem cells in patients embarking on melphalan-based therapy, even if transplant was not defi- nitively planned and perhaps more likely to treat transplant- eligible patients with non-melphalan-based approaches.Similar to that observed in multiple myeloma in older studies, we were able to demonstrate that the effect of melphalan on bone marrow reserve is dose-dependent. Estimated cumulative doses of melphalan of 150 mg or less, which is equivalent to a median of two cycles, were asso- ciated with higher stem cell yields and 96% success rate for goal stem cell collection as compared to higher doses, where the success rate was 85%. This difference in suc- cessful stem cell collection was not statistically significant in our analysis (p = 0.1), but this possibly reflects small sample size in the two melphalan dose groups. Our findings demonstrate that melphalan should be used cautiously prior to stem cell collection in patients with AL amyloidosis and that prolonged use should be avoided. Based on the results observed with dose effect of melphalan in our cohort, we suggest that duration of exposure before stem cell collection should be limited to no more than two cycles of therapy.Bortezomib in combination with cyclophosphamide was the most common non-melphalan-based regimen. Cyclo- phosphamide and melphalan are both alkylators, but appear to have a differential effect on stem cells. Cyclopho- sphamide exposure did not impact stem cell yields and patients in the melphalan group had lower stem cell yields compared to cyclophosphamide. Lenalidomide has also been shown to have an adverse impact on stem cell col- lection in multiple myeloma [18]. We observed a trend toward lower day-1 yield and average yield per collection and more apheresis sessions in lenalidomide-exposed patients compared to patients not exposed to any che- motherapy. While we cannot make a definitive conclusion on the impact of lenalidomide with our small sample size of27 patients, our findings suggest that lenalidomide may have an unfavorable impact on stem cell yields in AL amyloidosis as well.Upfront use of melphalan is declining in the United States [19]. However, melphalan still remains an important drug in the therapy of AL amyloidosis [20, 21] and con- tinues to be used as initial therapy in many parts of the world, either alone or in combination with bortezomib. In fact, a phase III clinical trial in Europe recently completed accrual comparing bortezomib–melphalan–dexamethasone vs. melphalan–dexamethasone combination in patients with amyloidosis (NCT01277016) [22]. Therefore, our results are relevant to the current management of patients with AL amyloidosis.

Our study has limitations due to its retrospective design and spanning of the study cohort over 15 years, as our practice of stem cell collection has evolved in that time period. To account for this, we carried out a sub-group analysis for patients who underwent collection after 2009, when a risk-adapted plerixafor algorithm was implemented at our institution. Similar results were observed in this sub- group analysis. In a small proportion of patients (13%), the dose of melphalan received prior to collection was not known, though the number of cycles of treatment was known. For these patients, the dose was estimated based on standard doses of melphalan used at our institution for the given regimen, and although not ideal, such imputation methods can be used in clinical studies of cancer patients to account for missing data [23, 24].

In conclusion, melphalan exposure adversely impacts bone marrow reserve and impairs stem cell yields in patients with AL amyloidosis compared to other chemotherapeutic agents or no chemotherapy exposure, even with the avail- ability of plerixafor and despite shorter duration of mel- phalan exposure compared to other regimens. We were able to demonstrate a dose–effect relationship of melphalan on stem cell yields and therefore recommend that duration of melphalan exposure prior to stem cell collection should be limited to as low as possible; ideally, not exceeding more than two cycles of
AMD3100 treatment.