Low-dose methotrexate in sickle-cell disease: a pilot study with rationale borrowed from rheumatoid arthritis

This was a prospective pilot study that enrolled young adult patients with sickle cell disease who were under chronic hydroxyurea treatment and had more than 3 severe VOC episodes/year, that were refractory to opioids for periods longer than 3 weeks duration. For inclusion purposes, a VOC episode was defined as any severe acute pain episode demanding emergency admission for opioid parenteral administration. Pregnancy and concomitant infection were exclusion criteria. Contraceptive measures were routinely advised for fertile female patients under chronic hydroxyurea. The study was conducted in the city of Campinas in the Brazilian Southeast region. The population in this metropolitan area is prevalently white but includes close to 25% individuals self-described as multiracial or black. About 230 SCD patients ≥18 years of age attend Boldrini Children’s Center each year. Recruitment for this study was opened for 10 months, during which 14 patients had severe disease and met the inclusion criteria. None of them declined to participate. Eleven carried HbSS and 3 had HbSC. There were 9 men and 5 women, with a median age of 23.5 years (range 18–32 years). Clinical and demographic data are presented in Table 1. All patients had experienced many more acute and severe pain crisis episodes than the inclusion threshold of 3/year, despite hydroxyurea treatment. One patient (#6) had 19 VOC episodes per month. The median number of VOC episodes in the other 13 patients was 3.3/month (95% CI 2.0–5.0). Seven patients presented avascular osteonecrosis of the femoral and/or humeral heads.

The experimental treatment consisted of one intramuscular injection of 10 mg of methotrexate per week for 12 weeks. A single weekly dose of 5 mg of leucovorin was administered orally 48 h after each methotrexate injection. Sickle cell disease is a condition characterized by hemolysis and increased erythropoiesis, which may compromise folate levels [7]. The rationale for leucovorin rescue in our protocol stems from the anti-folate activity of methotrexate. Leucovorin is a reduced form of folate that limits the potential toxic effects associated with the inhibition of folate-dependent enzymes by methotrexate. Interruption of ongoing treatment with hydroxyurea and opioids was not recommended. Hematological and biochemical parameters were routinely recorded.

The primary endpoint was the decrease in frequency and intensity of severe acute pain episodes that required emergency medical assistance during the 12-week long study, compared to an equivalent period preceding the MTX treatment.

The secondary endpoints were the blood levels of inflammatory markers and the quality of life (QOL), which were evaluated at three time points (0, 6 and 12 weeks) of MTX treatment. Each patient was considered as his/her own control. S.R.B, P.O.C.L., R.A., J.A.Y., and A.B.A.L. analyzed and had access to primary clinical data. This study followed the Declaration of Helsinki, was approved by the Institutional Ethics Committee, and was registered in the Brazilian National Ethics Database/Plataforma Brasil (CAAE: 36641414.9.0000.5376). All participants gave written informed consent. This study was registered at http://​www.​who.​int/​ictrp/​en/​ and http://​www.​ensaiosclinicos.​gov.​br (World Health Organization International Clinical Trials Registry Platform/Brazilian Clinical Trials Registry Identifier: RBR-2s9xvn).

The general effectiveness of methotrexate treatment was evaluated by the patients in a scale of 0–10 at the last follow-up visit. Pain quantification was performed with the Brazilian validation of the McGill Pain Questionnaire, which contains 68 pain descriptors [21, 22]. Considering that avascular osteonecrosis pain has a different course from acute painful crises, patients with that complication were asked to quantify separately the treatment efficiency on the associated chronic pain from 0 to 10. Chronic pain intensity was plotted on a scale, in which 0 represents no pain (Treatment efficiency: 10) and 1.0 represents the maximal baseline chronic pain (Treatment efficiency: 0).

Quality of life was assessed according to the Brazilian Portuguese validation of the SF-36 Questionnaire, which contains four 0–100 subscales for the physical and mental domains [23, 24]. Symptoms of depression were measured with the 21-item Beck inventory [25, 26].

Blood samples (20 ml) were collected in EDTA tubes before the first meal. Plasma aliquots were kept frozen at −70 °C until the end of the study, when they were analysed. Measurements of cytokines, C-X-C motif chemokines (CXCL), and adhesion molecules (IL-1β, IL-4, IL-6, IL-8, IL-10, IL-17, IFN-γ, TNF-α, sICAM1, sVCAM1, sE-Selectin, CXCL1, CXCL9, CXCL10, and CXCL12) were performed with Quantikine ELISA kits (R&D Systems, Inc., Minneapolis, MN, USA), according to the manufacturer’s instructions. sP-Selectin was measured with the human sP-Selectin/CD62P immunoassay kit from R&D Systems. Details regarding the sensitivity, specificity, linearity, and intra-/inter-assay precision are available in the manufacturer’s website (http://​www.​rndsystems.​com). White blood cell counts, hemoglobin F, reticulocytes, biochemical markers of kidney and liver function, and C-reactive protein were measured in freshly collected blood samples.

The number of VOC episodes during the 12-week methotrexate treatment was not significantly different from that recorded in a preceding trimester: the medians were 12 (95% CI 6–26) and 10.5 (95% CI 6–21), respectively, (P = 0.6240; Fig. 1a). However, patients were asked to rate the general effectiveness of MTX treatment in a scale of 0–10 at the last follow-up visit of the study: 8 out 14 individuals reported general improvement in their condition, with a score ≥5 (median 7, 95% CI 5–8.5, n = 8). Their effectiveness perception was supported by the lower score obtained with the McGill Pain Questionnaire at week 12, compared to the one administered at week 0. The median McGill Pain Index (MPI) dropped 18.9% at 12 weeks in these 8 patients, from 45 (95% CI 37–57) to 36.5 (95% CI 9–49) (P = 0.0469, n = 8, Fig. 1b). The Additional file 1: Table S1 shows the MPI results for all 14 participants by the end of the study. The median MPI decreased 27.2% from 49.5 (95% CI 37–57) to 36 (95% CI 9–45) in 8 patients, and increased 22.5% from 40 (95% CI 37–52) to 49 (95% CI 48–57) in 4 cases. This index was nearly unchanged (<3%) in 2 individuals. Overall, there was no significant difference (P = 0.931).

In addition to acute painful VOC episodes, patients may also experience nociceptive chronic pain caused by avascular osteonecrosis that is difficult to quench, is resistant to opioids and hardly ceases in-between crises [27]. Seven of the 14 patients who entered the study had avascular osteonecrosis of the femoral and/or humeral heads demonstrated by magnetic resonance imaging (Table 1). Five of them (#1, #3, #4, #8 and #9) referred a chronic pain reduction ≥50% after MTX treatment (P = 0.015). Only 2 patients (#6 and #7) did not perceive any benefit at week 12 (Fig. 1d).

The osteonecrosis pain data was combined with the MPI results, in order to compile a list of responders and nonresponders to MTX therapy, which revealed pain relief in 71% of the patients (10/14). This response was heterogeneous, probably reflecting the overlapping occurrence of acute and chronic pain (Table 2). Three patients (#1, #3 and #8) exhibited the highest relief because they had both lower MPI and decrease of osteonecrosis pain. Patient #1 evolved from having continuous hip pain, even between VOC episodes, to a situation in which there were periods without any pain.

Patient #8 had bilateral osteonecrosis of the femoral and humeral heads, with previous hip joint replacement on the right side. His clinical picture improvement during MTX therapy allowed him to walk without crutches.

It is interesting that there were two patients (#4 and #9) who did not experience pain reduction associated to acute VOC episodes, but had ≥50% decrease in chronic osteonecrosis pain. Patient #4 illustrates well this category of response: at the end of the study, her MPI was 19% higher. However, the chronic pain completely disappeared in her right shoulder and reduced 50% in her hips. She also regained mobility of the gleno-humeral joint that had been greatly compromised by the humeral head necrosis. Patient #9 had a similar MPI before and after MTX treatment, but had 50% reduction of the osteonecrosis pain intensity in his left hip joint.

To all 10 patients that were classified in Table 2 as MTX responders and two of those that fell into the MTX nonresponder category (#6 and #7), it was asked to define the longest uninterrupted period without pain at home. As indicated in Fig. 1c, the majority (10 individuals) responded that they achieved longer pain-free periods under MTX treatment, as compared to the trimester that preceded the study [median: 9 days (95% CI 1.5–21) vs 1 day (95% CI 0–5.0), P = 0.0020, n = 12]. Six of the participants said that they felt continuous persistent pain all the time before the MTX treatment, four of whom (#1, #3, #13 and #14) admitted gaining pain-free days under MTX therapy.

There was a transient drop in the scores of the SF-36 role emotional subscale at week 6 of the MTX treatment, but patients scored similarly when results from week 0 and 12 were compared (Additional file 1: Figure S1A–E). Among the 8 domain subscales of the questionnaire [23], only the one that measures physical functioning was significantly and persistently improved by the methotrexate treatment. In concordance with the pain reduction reported by most patients, a 44.4% median gain of physical function could be identified in the entire study group at week 12 (Additional file 1: Figure S1A). This finding was also evident in the MTX responder subgroup as defined in Table 2, which had a median score 105.8% higher by the end of the study (Additional file 1: Figure S1B). Patients with avascular osteonecrosis scored almost double of the initial level of physical functioning by the end of the MTX treatment [Median: 65 (95% CI 5–85) vs 33 (95% CI 0–45), P = 0.004, n = 7], while those without avascular osteonecrosis scored similarly at weeks 0 and 12 (P = 0.249, n = 7) (Fig. 1e).

There was no major change pattern related to depression in the studied patients as assessed by the Beck Inventory [25] (Additional file 1: Figure S2). Four individuals in the avascular osteonecrosis subgroup kept their depression status (#1, #7, #8 and #9), one moved from moderate to mild (#3), one moved from mild to moderate (#4), and one changed from minimal to mild (#6). The results were similar for the group without avascular osteonecrosis, with unaltered status in three individuals (#12, #13 and #14), improvement from mild to minimal depression in one case (#11), and change from minimal to mild in two occasions (#2 and #5) and from minimal to moderate in another (#10).

Table 3 summarizes the cytokine, adhesion molecule and chemokine plasma levels at weeks 0, 6 and 12 of the study. C−Reactive Protein (CRP) was also used as an additional unspecific inflammation marker in 9 out of the 14 patients, being moderately to highly increased in the blood of most individuals tested at week 0, except one (#9). There was a median drop in CRP concentration of 58.4% in six individuals (#7, #8, #11, #12, #13 and #14) and an increase in two patients (#6 and #10: 90.5 and 44%, respectively). The CRP levels remained close to normal baseline in one case (#9). The median CRP levels ranged from 0.84 mg/dL (week 0) to 0.55 mg/dL (week 12) (P = 0.359, Table 3).

Among the inflammatory markers tested, MTX treatment had a highly significant impact on two chemokines, that possess both chemotactic and angio-modulatory properties. The most significant effect was exerted on the pro-angiogenic factor CXCL12, with a plasma concentration upward trend registered in all patients (P < 0.0001) (Table 3; Fig. 2a). The increase of CXCL12 levels occurred significantly in MTX-responder patients (P = 0.002) and trended upward in nonresponders (P = 0.125), irrespectively of the presence of avascular osteonecrosis (P = 0.015; Fig. 2b, c). The chemokine CXCL10 concentration followed a similar pattern described above, with an upward trend induced by MTX treatment (P = 0.0463) (Table 3; Additional file 1: Figure S4B–D).

High plasma levels of IL-1β and IL-10 were recorded among all patients at the 3 study point determinations. There was no significant up- or down-regulation in plasma concentration of these two and twelve additional inflammatory proteins measured at weeks 0, 6, and 12 of the MTX treatment, including major cytokines, chemokines, and adhesion molecules: IL-1β, IL-4, IL-6, IL-8, IL-10, IL-17, IFN-γ, TNF-α, sICAM1, sVCAM1, sP-Selectin, sE-Selectin, CXCL1, and CXCL9. These results are illustrated in Table 3 alongside normal reference values. In the case of TNF-α, methotrexate induced a down-regulation of the cytokine plasma levels in 60% of the patients at week 12 of the study (8 out of 12), with a more pronounced decrease in the middle time point, from 28.4 pg/ml at week 0 to 5.8 pg/ml at week 6 (P = 0.0753) (Table 3; Additional file 1: Figure S4A).

Finally, there was no significant alteration in hematological and biochemical parameters tested during the study (Additional file 1: Figure S3).

Considering that all patients were in routine use of hydroxyurea, plus concomitant use of other drugs, such as opioids, antibiotics, chlorpromazine or carbamazepine, it was difficult to identify/separate any adverse effect related to MTX itself. According to the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events v3.0 (CTCAE) [29], three patients developed low blood platelet counts: grade 1 (Patient #5: 91,000/μl) and grade 4 (Patient #10: 24,000/μl and #13: 11,000/μl) attributable to hydroxyurea that returned to normal levels in a few days after discontinuation of that medication. All these three patients had previous history of hydroxyurea-induced hematological toxicity. Nevertheless, MTX treatment was interrupted for 1–2 weeks and resumed thereafter. Patient #2 had an episode of seizure (grade 2), possibly related to SCD. The following infectious events were observed: dengue fever (Patient #8), common cold (Patient #12), pharyngitis (Patient #11 and #3), sinusitis (Patient #12), pneumonia (Patient #8), and FUO (Patient #6). Patients received antibiotics when indicated and MTX administration was temporarily suspended. Patient #2 had a central venous line infection (grade 5). Although catheter and blood cultures became negative after antibiotic treatment, the infection reappeared later on, and the patient died of pneumonia and septic shock 11 days after the end of the study. In this last case there was a drop in WBC counts (grade 2: 2070/μl), platelets (grade 3: 28,000/μl), and lymphocyte counts (grade 3: 270/μl), being difficult to differentiate between peripheral sepsis consumption or additional myelosuppression.