Cyclophosphamide for salvage therapy of chronic graft-versus-host disease: a retrospective analysis

All patients treated with cyclo for cGvHD at the transplantation program of the University Hospital Regensburg between 01/2010 and 11/2019 (n = 13) were included in the retrospective analysis approved by the institutional review board (no. 19-1585-104). The National Institute of Health (NIH) consensus criteria grading for cGvHD and the immunosuppressive regimen were assessed at the time of the first cyclo administration and repeated 3, 6, and 12 months of therapy as part of clinical routine during regular follow-up outpatient visits [25]. No other immunosuppressive agents were started within 4 weeks before start of cyclo and response assessment was stopped at the switch to any additional new immunosuppressive treatment line. Patients receiving cyclo as part of treatment for relapsed multiple myeloma after alloHSCT were excluded. Intention to treat was steroid-refractory cGvHD in nine patients, whereas three patients suffered from cGvHD-associated (glomerulo-)nephritis and one patient from vasculitis-like CNS manifestation of cGvHD.

Response was defined as complete remission (CR) in case of resolution of all cGvHD manifestations. Partial remission (PR) described an improvement of at least one organ grade or in case of nephritis decreased proteinuria (mg protein/g creatinine) > 50% without an increase in creatinine levels > 20%. Mixed response (MR) was defined as complete or partial remission in at least one but progression in another organ site, whereas progressive disease (PD) meant a progression without any improvements. No change in organ grading or cerebral lesions was classified as stable disease (SD).

For assessment of infectious complications, the common terminology criteria for adverse events version 5.0 (CTCAE 5.0) were applied. Toxicities with a CTCAE grades I and II were not captured in the analysis.

We identified and analyzed 13 patients (male n = 8, female n = 5) treated with cyclo between 01/2010 and 11/2019 for cGvHD. Initial diagnoses leading to alloHSCT were myeloid disorders (acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), myeloproliferative neoplasia (MPN)) in eight patients, whereas five patients suffered from lymphatic malignancies (multiple myeloma (MM), non-Hodgkin lymphoma (NHL)). All except one patient received a matched donor graft of related (n = 7) or unrelated (n = 6) donors. The predominant graft source was peripheral blood stem cells (PBSC, 12/13), while one patient received a bone marrow graft. GvHD prophylaxis consisted of a calcineurin inhibitor (ciclosporin A (CsA) or tacrolimus) and methotrexate (MTX) with (n = 8) or without ATG (n = 4). Only one patient received a prophylaxis with CsA and mycophenolate mofetil (MMF). Ten of 13 patients developed acute GvHD (aGvHD) on median day 27 (range: 13–63, no assessment in one patient) post-alloHSCT with a maximum grade (by Glucksberg criteria) of 2 or higher in 46%. Patient characteristics and history of aGvHD are summarized in Table 1.

Onset of cGvHD was quiescent in eight, de novo in three, and progressive in two patients. The onset of cGvHD was observed on median day 202 (range: 84–571) upon alloHSCT; thrombocyte count at onset was < 100/nl in five patients (no assessment available in one patient). A median of three (range 1–6) organs or sites were involved at start of treatment: eyes (6/13), lung (6/13), skin (4/13), fascia and joints (4/13), mouth (3/13), kidney (3/13), liver (1/13), genital (1/13), gastrointestinal (1/13), and CNS (1/13). Cyclo was started on median day 938 (range 218–8282) after alloHSCT and on day 509 (range 42–8193) after cGvHD onset, respectively. Prior to cyclo, a median of 2 therapy lines (0–8) had been administered. The individual patient’s treatments and the respective indication for use of cyclophosphamide are listed in Table 2. At the start of cyclo therapy, GvHD grading was moderate in three and severe in six patients. Three patients suffered from GvHD-associated (glomerulo-)nephritis with proteinuria and one patient from ischemic CNS lesions with focal neurologic deficits due to vasculitis-like cerebral cGvHD manifestation. Twelve patients received a continuous oral dose of 50 mg/day (n = 6) or every 2–3 days due to impaired blood counts (n = 6), whereas one patient was treated with a pulse therapy (initial 4 applications fortnightly, following 3 applications monthly, each 7.5 mg/kg intravenously). Steroid dose at start of cyclo treatment was in median 0.295 mg/kg (range: 0.11–1.18 mg/kg). The duration of cyclo therapy was median 153 days (range: 14–486), and therapy was still ongoing in two patients at the time of last assessment (date 11/30/2019). Median follow-up after treatment was 407 days (range: 86–1534). A more detailed overview is given in Table 3.

The 3-month follow-up was reached by nine patients with one patient achieving complete remission and three patients partial remission resulting in an overall response rate (ORR) of 31%. Because of a sustained and progressive renal improvement for longer than 6 months, one of these patients was classified as partial remission despite a 48.8% (instead of 50%) decrease in proteinuria. In contrast, three patients had a stable disease and two further patients showed a progressive disease. New immunosuppressive medication was started in four cases precluding any response assessment. Of those, one patient received mesenchymal stem cells (MSCs) in a status of partial remission 1 month after cyclo start and three patients had aggravating cGvHD with pulmonary manifestations in two cases (Table 3). None of the patients suffered from relapse of the underlying malignancy; the failure-free survival (FFS) was 69% after 3 months. Within the first 3 months, immunosuppressants besides cyclo were reduced in four patients.

At 6 months after initiation of the cyclo treatment, one patient still had a complete remission and two patients showed partial remission, whereas one patient remained unresponsive. One additional patient developed progressive disease (n = 3) with new oral and aggravated ocular cGvHD. No further patient needed a new ISM (Table 3). After 6 months, ORR was 23% and FFS was 69%.

At 12-month follow-up, a total of six patients received new ISM and three patients have not yet reached the point of time. Two patients still showed a partial remission, one patient had a stable disease and one patient had progressive disease, displaying an ORR of 15% and an FFS of 54% (Table 3).

Cyclo is a commonly used treatment option in vasculitis. Therefore, we analyzed patients receiving cyclo due to cGvHD-associated vessel diseases (cGvHD-associated nephritis, cGvHD-related PML). Three patients showed a cGvHD-associated GN and nephritis, respectively, displayed by increased creatinine levels (median: 119 μmol/l, range: 81–225) and proteinuria (median: 2799 mg/g creatinine, range: 1669–3659). The subtypes were classified as membranous GN (n = 1) and focal segmental GN (n = 1) by biopsy and as tubulointerstitial nephritis by urine examination (n = 1). During the course of cyclo treatment, all patients showed a partial remission of proteinuria (mean of best reduction: 81.7%) with concomitantly improved renal function within 3 months after first cyclo administration (Fig. 1). Of note, one patient required new ISM due to newly developed moderate lung GvHD during treatment and one patient showed increasing proteinuria 1 month after discontinuation of cyclo therapy. All patients suffering from renal manifestation of cGvHD received angiotensin-converting enzyme (ACE) inhibitors or angiotensin II receptor blockers (ARBs), respectively. However, treatment was started before cGvHD onset due to arterial hypertension in all patients.

The patient suffering from cGvHD-associated ischemic cerebral lesions showed only a short phase of stable disease. Due to clinically significant adverse events with headache, alopecia, arthritis, and nausea, cyclo had to be discontinued after 2 months. One month later, the patient developed ocular cGvHD; 3 months later, ischemic lesions progressed and immunosuppressive therapy with methotrexate was started. Thus, we considered the patient to be a non-responder.

An early neutropenia was observed in two patients receiving cyclo orally after 1 and 4 weeks, respectively. Four patients were hospitalized (CTCAE grade III) due to an infection on median day 52 (range: 11–143) after start of cyclo treatment, of whom three suffered from respiratory symptoms and one from a urinary tract infection. None of the patients developed grade IV (intensive care unit) or V (death) infectious complications. The listed infections occurred during or immediately after cyclo treatment, and only one patient had meanwhile received further immunosuppressive treatment with mesenchymal stem cells. A correlation between response and infectious complications was not detectable, considering the low patient numbers.

As shown in Fig. 2, steroid dose of all patients reaching the follow-up was significantly reduced during the first 3 months of therapy by a median of 75% (range 0–100%). After 6 and 12 months, required steroid doses were increasing compared with the 3-month follow-up, but still reduced by 58 and 44% as compared with the initial dose. However, after 6 months, one patient received a tripled dose of steroids due to cGvHD progression, whereas all other patients had reduced or stable steroid need (Table 3).

Cyclo poses an uncommon cGvHD therapy and a predictive marker supported selection of suitable patients simplifies the decision-making process. Moreover, the number of patients precludes any valid statistics. Patients achieving complete or partial remission were characterized by an advanced age (mean 61.5 vs 50 years), lower number of pre-cyclo treatments (mean 1 vs 4.6), no further non-steroid ISM at cyclo start (0 vs 1.3), a shorter period until cyclo treatment (mean 281 vs 1755 days), and a lower platelet count at cGvHD onset (mean 99 vs 173/nl). However, cyclo and steroid dose did not predict response. In contrast, patients developing adverse events (CTCAE ≥ III) during or after cyclo treatment were younger (mean 44.5 vs 57.6 years), showed a higher number of pre-cyclo treatment lines (mean 4.3 vs 3.1), increased non-steroid ISM at cyclo start (mean 1.5 vs 0.7), and elevated steroid doses at cyclo start (0.69 vs 0.22 mg/kg bodyweight). In our cohort, we did not find a dependency of the cyclo dose on adverse events.