Autologous haematopoietic stem cell transplantation affects long-term progression independent of relapse activity in aggressive multiple sclerosis: a comparative matched study

Abstract

Background

Treatment of progression independent of relapse activity (PIRA) is a relevant unmet need in multiple sclerosis (MS), being only modestly affected by disease-modifying treatments (DMTs) which target predominantly adaptive immunity in the periphery. As chemotherapy administered during autologous haematopoietic stem cell transplantation (AHSCT) is bioavailable within the central nervous system (CNS), the hypothesis that AHSCT could affect long-term PIRA was explored in aggressive relapsing-remitting (RR)-MS.

Methods

Retrospective propensity-score matched study including RR-MS patients who received BEAM/ATG AHSCT or started natalizumab (NTZ, ie, controls) at our centre in Florence in the period 2007–2018. Main outcome: cumulative proportion of patients with PIRA during NTZ treatment epoch (ie, censoring controls at NTZ discontinuation) and whole follow-up (NTZ-other[o]DMTs, that is, including switch from NTZ to alternative DMTs).

Results

Thirty RR-MS were included in each group; median follow-up duration was 106 (6–209) months. NTZ was discontinued by 29/30 patients, who subsequently started alternative DMTs. Cumulative proportion of patients with PIRA did not differ between the two groups during the NTZ treatment epoch (p=0.990), but it was lower in AHSCT-treated compared with NTZ-oDMTs treated patients over the whole follow-up, being 10% vs 21% at year 5, and 10% versus 49% at year 10, respectively (p=0.020). AHSCT was superior to NTZ on relapses and NEDA-3, and to NTZ-oDMTs on all the secondary outcomes analysed. Baseline age and Expanded Disability Status Scale independently predicted PIRA in the whole cohort.

Conclusion

Timely treatment with AHSCT or DMTs targeting inflammation in both the peripheral and CNS compartments might prevent long-term PIRA in aggressive RR-MS.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Autologous haematopoietic stem cell transplantation (AHSCT) is highly effective on new focal inflammation in relapsing-remitting multiple sclerosis (RR-MS), but its effect on progression independent of relapse activity (PIRA) has been poorly investigated, and no comparative data with disease-modifying therapies (DMTs) are available to date.

WHAT THIS STUDY ADDS

• Treatment with AHSCT reduced the risk for long-term PIRA in aggressive RR-MS compared with DMTs, and provided higher rates of sustained remission of MS activity. Exploratory data suggest that AHSCT could also delay the transition to secondary-progressive MS.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Switching to AHSCT or DMTs targeting inflammation in both the peripheral and CNS compartments might dramatically change long-term disability trajectories in people with aggressive MS, providing that this is performed before the individual neurologic reserve is exhausted.

Introduction

Multiple sclerosis (MS) is a chronic inflammatory and degenerative disease of the central nervous system (CNS) where axonal damage beginning even prior to clinical onset¹ may be perpetrated over time by overt and hidden pathogenetic drivers, ultimately determining irreversible disability accrual once brain reserve is exhausted.² On clinical grounds, people with MS may accumulate disability either as incomplete recovery from a relapse or independently from relapses, defined as relapse-associated worsening (RAW) and progression independent of relapse activity (PIRA), respectively.³ PIRA is the main driver of disability accrual in progressive MS,⁴ although it may occur in early relapsing-remitting (RR)-MS,⁵ and it accounted for 80%–90% of the overall disability events in patients receiving ocrelizumab in the OPERA trials.³ Early use of high-efficacy (HE-) disease-modifying treatments (DMTs) reduces the risk of long-term disability accrual compared with delayed use,⁶ but this seems to be mediated mostly by a reduction of RAW, suggesting that PIRA is prompted by different pathogenetic mechanisms.⁴ These likely include inflammation compartmentalised within the CNS and degenerative processes independent from inflammation, which plausibly contribute to PIRA at a different extent across individuals, and within each individual over different phases of the disease.⁷ The ability of a treatment to cross the blood-brain barrier (BBB) would be a prerequisite for acting on compartmentalised inflammation, but only a few DMTs are bioavailable within the CNS.⁸ Conversely, chemotherapy drugs crossing the BBB⁹ are administered during autologous haematopoietic stem cell transplantation (AHSCT), a haematological procedure endorsed as a standard of care for the treatment of aggressive RR-MS failing HE-DMTs.¹⁰ The hypothesis that AHSCT with BEAM/ATG protocol could affect PIRA in RR-MS was therefore explored, over median 8.8 years of follow-up, by comparing AHSCT-treated cases with matched control patients (CTRL) who, in the same index period, started treatment with natalizumab (NTZ), a HE-DMT reducing lymphocyte trafficking across the BBB.¹¹

Methods

Patient population and procedures

All the RR-MS patients diagnosed according to the McDonald criteria¹² who had been consecutively enrolled in an open-label monocentric study on AHSCT in MS or who started treatment with NTZ (for at least 6 months) at the Neurology 2 Department of the Careggi University Hospital in Florence (Italy) in the index period 2007–2018 were considered eligible as potential cases and CTRL, respectively.

The start- and end-year of the index period were chosen based on the dates in which NTZ (December 2006) and ocrelizumab (September 2018), respectively, were approved for reimbursement for the treatment of MS in Italy, in order to allow the inclusion of patients treated with either NTZ or AHSCT over a similar epoch, with no major changes in eligibility to NTZ treatment.

Patients were approved for treatment with AHSCT after individual case discussion within multidisciplinary meetings including at least two neurologists and haematologists of the Centre, based on current eligibility criteria for AHSCT, detailed in online supplemental material. AHSCT was performed at the Cellular Therapies and Transfusion Medicine Unit of the Hospital, in collaboration with the Tuscan Region MS Referral Centre. The conditioning protocol used was the intermediate-intensity regimen BEAM (carmustine, etoposide, cytosine arabinoside and melphalan) plus rabbit anti-thymocyte globulin (ATG; total dose 7.5 mg/kg); haematopoietic stem cells (HSCs) were mobilised with IV cyclophosphamide (4 g/m² body surface area) and granulocyte colony-stimulating factor (G-CSF; 10 µg/kg per day), as previously reported.¹³ Supportive therapies and infection prophylaxis were administered according to local protocols.

NTZ was prescribed, according to the Italian reimbursement criteria, to RR-MS patients with highly active disease failing DMTs, or treatment-naïve with aggressive onset. NTZ 300 mg was administered at the Tuscan Region MS Referral Centre of the Hospital, with IV infusions every four to six weeks according to clinical practice, depending on John Cunningham virus (JCV) status.

Study baseline was defined as mobilisation of HSCs or first NTZ administration for the AHSCT and CTRL groups, respectively. Neurological follow-up with Expanded Disability Status Scale (EDSS)¹⁴ assessment was carried out according to standardised clinical practice, with scheduled evaluations every 6 to 12 months performed by a pool of neurologists expert in MS, trained on the harmonized procedures of the MS Referral Centre.

Study design and aims

Retrospective monocentric matched study aimed at exploring differences in the cumulative proportion of PIRA between RR-MS patients who received AHSCT (cases) or NTZ followed by other (o)DMTs (CTRL). Secondary outcomes included all the following: cumulative proportion of patients with RAW, overall EDSS worsening, relapse, and no evidence of clinical and radiological disease activity (NEDA-3) survival. Conversion to secondary-progressive (SP-)MS and predictors of PIRA were investigated as exploratory outcomes.

Outcome measures

EDSS worsening was defined as a confirmed increase of ≥1.5, 1.0 or 0.5 EDSS points if baseline EDSS was 0,<5.5 or ≥5.5, respectively. EDSS worsening was further classified into: (i) RAW, if occurring ≤90 days after or ≤30 days before the onset of a relapse, or (ii) PIRA, if occurring >90 days after and >30 days before the onset of a relapse. A relapse was defined as new-onset or abrupt worsening of neurological symptoms attributable to MS, lasting at least 24 hours, in the absence of fever or concomitant infection, associated with an objective modification at neurological examination. NEDA-3 was defined as absence of all the following: relapses, EDSS worsening, MRI activity (new T2 lesions compared with a re-baseline scan taken on average 3–6 months after study baseline, or gadolinium-enhancing lesions at any time).

Safety outcomes included secondary autoimmunity, neoplasms and deaths up to the last follow-up.

Statistical methods

Patients were selected using 1:1 propensity score (PS) matching without replacement from two cohorts including all the RR-MS patients who had either started treatment with NTZ (n=76) or received BEAM/ATG AHSCT (n=33) at our centre during the index period (online supplemental table 1). Thirty out of 33 AHSCT-treated RR-MS patients were matched to 30 out of 76 NTZ-treated RR-MS patients adopting the SPSS Python extension Fuzzy with a tolerance level of 0.10. The matching was blind to the clinical information related to the post-treatment outcome variables, and it included as confounders sex, age, disease duration and EDSS at baseline.

Continuous and dichotomous variables are summarised as median (range) or mean (SD) based on data distribution, and number (frequency), respectively. Comparisons between the groups with Mann-Whitney/T-test (as appropriate for data distribution) or χ² test for continuous and dichotomous variables, respectively. Survival free from the index event was estimated by Kaplan–Meier analyses and compared between groups with Log-rank test. Predictors of PIRA were explored by a Cox regression model including baseline age, EDSS, sex and number of prior DMTs.

As discontinuation of treatment was observed in the CTRL group only, following the ICH E9(R1)-ADDENDUM,¹⁵ the outcomes in this group were analysed considering (i) the NTZ treatment epoch, that is, censoring patients at the time of last NTZ administration (hypothetical strategy; CTRL NTZ), and (ii) the whole follow-up period after the first dose irrespective of treatment discontinuation, that is, including treatment with oDMTs after NTZ withdrawal, hereafter defined as CTRL NTZ-oDMTs (ITT/policy strategy). Study design is exemplified in figure 1. Pairwise censoring was applied to account for disproportionate follow-up between groups during the NTZ treatment epoch and used to perform a sensitivity analysis on the primary outcome over the whole follow-up. Patients who received AHSCT after NTZ discontinuation were censored at the date of AHSCT. The software programme used for statistics was SPSS V.25 for Windows; graphing with SPSS and Origin. A two-tailed p value <0.05 was considered significant.

Figure 1.

Study design. Thirty patients who received autologous haematopoietic stem cell transplantation (AHSCT) and 30 control (CTRL) patients were selected adopting a 1:1 propensity score (PS) matching from two cohorts of 33 and 76 RR-MS patients including all the patients who had received AHSCT or started treatment with natalizumab (NTZ) at our centre in the index period, respectively. Study baseline was defined as mobilisation of haematopoietic stem cells and first NTZ administration for patients in the AHSCT and CTRL group, respectively. As AHSCT is a one-off treatment, no disease-modifying therapy (DMT) was routinely administered after transplant to patients in the AHSCT group, who therefore entered a period of clinical and MRI follow-up. Patients in the CTRL group received NTZ 300 mg intravenously every 4 to 6 weeks and discontinued NTZ treatment according to clinical practice, followed by treatment with other (o)DMTs up to last follow-up. Outcomes were analysed both during the NTZ treatment epoch (ie, censoring CTRLs at NTZ discontinuation and adopting pairwise censoring for disproportionate follow-up between groups) and the whole follow-up (ie, from baseline to the latest clinical assessment), including switch from NTZ to oDMTs.

Standard protocols approval and patient consent

The study protocol was approved by the local ethics committee (Comitato Etico Area Vasta Centro, Regione Toscana; approval number 17696_oss); written informed consent was collected by participants according to local regulations.

Results

Patient characteristics

Baseline characteristics of the 60 patients included were overall balanced (table 1), except for a higher number of previous DMTs in the AHSCT compared with the CTRL group (p=0.001), being at least 4 DMTs received by 23% and 3% of the cases, respectively. Cumulative exposure to DMTs before the study baseline is reported in figure 2. HE-DMTs were overrepresented in the AHSCT group, where 13/30 (43%) patients had previously received NTZ for a median of 28 (23–57) doses; the main reasons for NTZ discontinuation were JCV+ (54%) and inefficacy (38%), being MS relapse experienced by 5 patients (associated with EDSS worsening in two cases); EDSS worsening alone was observed in two further patients. The median interval between NTZ and AHSCT was 9 (6–38) months.

Table 1.

Baseline clinical-demographic characteristics of the RR-MS patients included in the AHSCT and CTRL groups

	AHSCT (n=30)		CTRL (n=30)		P value
	Median	(Range)	Median	(Range)
Age, y	35	(20 – 53)	37.5	(18 – 63)	0.277
Disease duration, y	9.5	(1 – 22)	10.5	(0 – 26)	0.657
Previous DMTs duration, y	6	(0 – 21)	6.5	(0 – 15)	0.463
Number of previous DMTs	3	(0 – 6)	2	(0 – 5)	0.001
ARR in the previous 2 y*	1.30	(0.96 – 1.64)	1.21	(0.95 – 1.45)	0.592
EDSS at baseline	2.75	(1.0 – 7.0)	2.75	(0.0 – 6.5)	0.982
Delta-EDSS in the previous y	0.5	(-1.5 – 1.5)	0	(-1.5 – 2)	0.439

	n	(Proportion)	n	(Proportion)	P value
Gender, female	22	(73%)	23	(77%)	0.500
Treatment-naïve	1	(3%)	2	(7%)	1.000

*Reported as mean (95% confidence interval, CI).

AHSCT, autologous haematopoietic stem cell transplantation; ARR, annualised relapse rate; DMT, disease-modifying treatment; EDSS, Expanded Disability Status Scale; RR-MS, relapsing-remitting multiple sclerosis; y, years.

Figure 2.

Cumulative exposure (in months) to disease-modifying treatments (DMTs) before the study baseline in individual RR-MS patients from the AHSCT (n=30; A) and CTRL (n=30; B) groups. A higher number of DMTs was received by patients in the AHSCT group compared with those in the CTRL group (median 3 vs 2, respectively; p=0.001); one and two patients from the AHSCT and CTRL group, respectively, were treatment-naïve. Notably, 13/30 (43%) patients from the AHSCT group had previously received NTZ (red bars). AHSCT, autologous haematopoietic stem cell transplantation; CTRL, controls.

Median pairwise censored follow-up in the NTZ treatment epoch was 29 months (range 6–64); median whole follow-up (ie, from baseline to the last assessment) was 106 (6–209) months, being shorter in the AHSCT (median 86.5 months; range 36–209) compared with the CTRL group (median 139 months; range 6–187); p=0.001.

NTZ discontinuation and treatment with other DMTs in the CTRL group

Twenty-nine/30 (97%) patients in the CTRL group discontinued NTZ after a median of 24 (6–59) administrations. The main reason for NTZ withdrawal was JCV+ (n=23; 79.3%), followed by inefficacy (n=2; 7%), adverse events (n=1; 3.4% - persistent eosinophilia) and other (n=3; 10.3% - including one each of the following: anti-NTZ antibodies, pregnancy and voluntary discontinuation).

Treatment with other DMTs over follow-up was started in 27 cases after a median of 3 (1–13) months. The first DMT administered after NTZ discontinuation was a moderate to HE-DMT in 20 cases (11 fingolimod, six cyclophosphamide, two rituximab and one alemtuzumab) and a platform therapy in the remaining seven (five interferons, one dimethyl-fumarate and one glatiramer-acetate). Sixty-three percent and 30% of the patients switched DMT twice or thrice over follow-up, respectively. Two patients eventually received AHSCT.

PIRA and other disability outcomes

During the NTZ treatment epoch, the cumulative proportion of patients with PIRA did not differ between the AHSCT and CTRL NTZ groups, being at years 2–3 of 4% and 4%, respectively (p=0.990; figure 3A). Up to the last follow-up, a lower proportion of AHSCT-treated patients experienced PIRA compared with those who received NTZ-oDMTs, being 10% versus 21% at year 5, and 10% versus 49% at year 10, respectively (p=0.020; figure 3B). This result was confirmed by a sensitivity analysis adopting pairwise censoring also during the whole follow-up (p=0.003; online supplemental figure 1). Cumulative proportion of PIRA events in the CTRL NTZ-oDMTs group was similar between patients with de-escalation (n=9, including platform therapies and no treatment) and those who switched to moderate/HE- DMTs (n=20), although the small sample size prevents drawing any conclusions (online supplemental figure 2).

Figure 3.

Disability outcomes in the AHSCT and CTRL groups over the NTZ treatment epoch (NTZ, ie, censoring CTRLs at NTZ discontinuation; A, C, E) and the whole follow-up including treatment with other (o)DMTs following NTZ withdrawal (NTZ-oDMTs; B, D, F). The cumulative proportion of cases with progression independent of relapse activity (PIRA; A, B), relapse-associated worsening (RAW; C, D) and EDSS worsening (E, F) did not differ between AHSCT and CTRL patients during the NTZ treatment epoch, but AHSCT was superior to NTZ-oDMTs in all the disability outcomes over the whole follow-up. The number of patients in observation at each timepoint is reported below each chart. AHSCT, autologous haematopoietic stem cell transplantation; CTRL, controls; DMTs, disease-modifying therapies; EDSS, Expanded Disability Status Scale; NTZ, natalizumab.

RAW (figure 3C,D) was observed exclusively in the CTRL group (4% at year 2; p=0.157), being cumulative RAW over follow-up significantly lower in AHSCT- (0% up to last follow-up) compared with NTZ-oDMTs-treated patients (21% at year 5 and 32% at year 10); p=0.002. The proportion of patients with cumulative EDSS worsening was similar between the two groups during the NTZ treatment epoch (at year 2: 4% AHSCT vs 8% CTRL NTZ; p=0.399, figure 3E), but over the whole follow-up, it was lower in the AHSCT (10% at years 5–10) compared with the CTRL NTZ-oDMTs group (38% and 65% at 5 and 10 years, respectively); p<0.001 (figure 3F).

Kaplan-Meier estimates of the efficacy outcomes analysed over the NTZ treatment epoch and whole follow-up are summarised in table 2.

Table 2.

Kaplan-Meier estimates of cumulative proportion of RR-MS patients with PIRA, RAW, EDSS worsening, relapses and NEDA-3 survival for the AHSCT and CTRL groups by year during the NTZ treatment epoch and whole follow-up period

NTZ treatment epoch: AHSCT/CTRL NTZ
Y/outcome	PIRA	RAW	EDSS worsening	Relapses	NEDA-3
1	0%/0%	0%/4%	0%/4%	0%/18%	100%/86%
2	4%/4%	0%/4%	4%/8%	0%/31%	96%/72%
3	4%/4%	0%/10%	4%/14%	0%/38%	96%/65%
p value	0.990	0.157	0.399	0.001	0.027

Whole follow-up: AHSCT/CTRL NTZ-oDMTs
Y/outcome	PIRA	RAW	EDSS worsening	Relapses	NEDA-3
1	0%/0%	0%/3%	0%/3%	0%/24%	100%/76%
2	7%/3%	0%/3%	7%/7%	0%/41%	93%/59%
3	7%/7%	0%/14%	7%/21%	0%/62%	93%/31%
4	10%/10%	0%/21%	10%/31%	0%/65%	90%/24%
5	10%/21%	0%/21%	10%/38%	0%/73%	90%/16%
6	10%/28%	0%/24%	10%/46%	0%/77%	90%/8%
7	10%/32%	0%/28%	10%/53%	0%/81%	90%/0%
8	10%/40%	0%/32%	10%/60%	7%/81%	76%/0%
9	10%/44%	0%/32%	10%/65%	7%/81%	76%/0%
10	10%/49%	0%/32%	10%/65%	7%/81%	76%/0%
p value	0.020	0.002	<0.001	<0.0001	<0.0001

AHSCT, autologous haematopoietic stem cell transplantation; CTRL, controls; EDSS, Expanded Disability Status Scale; NEDA-3, no evidence of disease activity-3 (ie, no relapses, EDSS worsening, MRI activity); NTZ, natalizumab; oDMTs, other disease-modifying treatments; PIRA, progression independent of relapse activity; RAW, relapse-associated worsening; RR-MS, relapsing-remitting multiple sclerosis; Y, years.

Relapses and NEDA-3

The cumulative proportion of patients with relapse was lower in the AHSCT compared with the CTRL group during both the NTZ treatment epoch and whole follow-up, being 0% vs 31% at year 2 (figure 4A; p=0.001), and 0% vs 73% at year 5 (figure 4B; p<0.0001), respectively. NEDA-3 survival at year 2 was 96% in the AHSCT group and 72% in the CTRL NTZ group (p=0.027; figure 4C). Over the whole follow-up, 90% and 76% of AHSCT-treated patients were NEDA-3 at years 5 and 10, respectively, compared with 16% at year 5, and 0% at years 7–10 of patients in the CTRL NTZ-oDMTs group (p<0.0001; figure 4D).

Figure 4.

Cumulative proportion of cases with relapses (A, B) and no evidence of clinical and radiological disease activity (NEDA-3; C, D) in the AHSCT and CTRL groups during the treatment epoch (NTZ, ie, censoring CTRLs at NTZ discontinuation; A, C) and the whole follow-up including treatment with other (o)DMTs following NTZ withdrawal (NTZ-oDMTs; B, D). AHSCT was superior to both NTZ and NTZ-oDMTs in suppressing relapses and disease activity. The number of patients in observation at each timepoint is reported below each chart. AHSCT, autologous haematopoietic stem cell transplantation; CTRL, controls; DMTs, disease-modifying therapies; NTZ, natalizumab.

Exploratory outcomes

Two out of 30 (7%) cases from the AHSCT group and 12/30 (40%) cases from the CTRL NTZ-oDMTs group converted to SP-MS up to the last follow-up, after a median of 16 and 21 years from MS onset, respectively. The cumulative probability of conversion to SP-MS at years 5 and 10 was 7% in the AHSCT group versus 21% and 32%, respectively, in the CTRL NTZ-oDMTs group (p=0.023; data not shown).

In the whole population, PIRA events were independently predicted by EDSS and age at baseline with an HR of 1.58 (95% CI 1.19 to 2.10; p=0.002) and 1.11 (95% CI 1.04 to 1.19; p=0.002), respectively.

Safety outcomes

No fatalities were observed in the AHSCT group. Two patients with advanced MS died over follow-up in the CTRL group. One thyroid cancer and one atypical breast ductal hyperplasia requiring surgery were diagnosed in two patients from the CTRL group. Secondary autoimmunity was observed in five (all thyroiditis) and four (three thyroiditis and one psoriasis) patients from the AHSCT and CTRL groups, respectively. One patient from the AHSCT group experienced avascular necrosis of the femoral bone.

Discussion

AHSCT induces a persistent and radical suppression of new focal inflammatory activity mediated by a renewal of the adaptive immune system,¹⁶ but it could also affect chronic inflammation and myeloid/microglial activation within the CNS.^{17 18} In recipients of allogeneic transplantation, up to 20% of microglial cells were replaced with donors’ bone-marrow derived macrophages, to an extent directly related to conditioning intensity, number of transplantations received and duration of post-transplantation survival.¹⁹ HSCT with busulfan conditioning induced microglial senescence in mice, followed by a gradual decrease to a critical microglial density which provided a permissive niche for engraftment of donor’s peripheral macrophages in the host brain.²⁰ In experimental autoimmune encephalomyelitis, bone marrow transplantation globally impacted CNS myeloid cells not only by integrating donor-derived myeloid cells, but also by altering the distribution and morphology of endogenous microglia, overall showing an increase in the neuroprotective myeloid state compared with pre-transplantation; interestingly, neuroprotective effects further improved when enhancing myeloid cell incorporation after a modified transplant.²¹ AHSCT could also affect innate immunity by acting on the lymphopoiesis/myelopoiesis ratio in the bone marrow, where it could revert a skewing towards an increased myeloid proliferation which was described in MS patients compared with healthy controls.²²

Based on this rationale and the putative contribution of pro-inflammatory microglia to PIRA,²³ AHSCT holds the potential to affect this phenomenon. We therefore explored whether AHSCT could halt PIRA in aggressive RR-MS by comparing disability outcomes in 30 AHSCT-treated patients 1:1 matched with CTRLs who, in the same index period, started treatment with NTZ. This was chosen as an inclusion criterion for the CTRL group for the following reasons: (i) NTZ was the most effective DMT available over the index period; (ii) based on the Italian reimbursement criteria, NTZ was prescribed to patients similar, as much as possible, to those generally considered eligible for AHSCT; (iii) it has a low likelihood of directly targeting chronic inflammation due to low BBB penetration,⁸ hence serving as optimal ‘negative control’ for analysing the potential effectiveness of AHSCT within the CNS compartment; and (iv) it eases the identification of PIRA thanks to suppression of relapses and, consequently, RAW. Almost all the CTRLs discontinued treatment with NTZ during follow-up, mostly due to progressive multifocal leukoencephalopathy risk, likely perceived as overwhelmingly high in an epoch when data on safety and effectiveness of extended dosing interval were not available yet.²⁴

The proportion of patients with PIRA, RAW and cumulative EDSS worsening was similar between the two groups during the NTZ treatment epoch, lasting for a median of 29 months. Over follow-up periods of median 1.5 to 3.5 years, disability accrual did not differ between AHSCT and matched fingolimod-, ocrelizumab- or alemtuzumab-treated RR-MS patients,²⁵ nor between PS-overlap-weighted AHSCT- and alemtuzumab- or ocrelizumab-treated RR-MS cohorts.²⁶ Similarly, at a median of 1.98 years of follow-up, no differences in disability accrual were reported between NTZ and AHSCT in a progressive MS population.²⁷ NTZ showed indeed some effectiveness in progressive MS in the ASCEND trial, where it reduced by 44% the risk for progression on the 9-Hole Peg Test compared with placebo and met the primary endpoint (composite disability outcome measure) in the 2-year open-label extension, although this was failed over 96 weeks.²⁸

The lack of differences between groups observed in our study during the NTZ treatment epoch could hence be due to some effect of NTZ on disability progression, and/or to a short observation period. It could therefore be questioned whether this would have persisted if NTZ treatment had been maintained over the whole follow-up. However, such a scenario seems unlikely from indirect comparisons with long-term data on NTZ treatment, as cumulative proportion of AHSCT-treated patients with EDSS worsening in our study (10% at year 10) corresponded to roughly one-third that reported in the TOP (27.8% at year 10)²⁹ and TYSTEN cohorts (34% at a mean follow-up of 97 months).³⁰ On the other hand, the cumulative proportion of patients with EDSS worsening in the CTRL NTZ-oDMTs group (65% at year 10) was almost double that observed in the TOP and TYSTEN studies, a figure that could be explained, at least in part, by a potential selection bias towards a more aggressive disease course and heterogeneity in treatment strategies after NTZ withdrawal, including de-escalation in some patients. Although the cumulative proportion of PIRA seemed to be similar between patients who de-escalated and those who did not, the small sample size in this subgroup analysis prevents us from dismissing the hypothesis that de-escalation could have contributed to poorer long-term outcomes in the CTRL group.

Over the whole follow-up, AHSCT was remarkably superior to CTRL NTZ-oDMTs on PIRA, and this was the only driver of disability accrual in AHSCT-treated patients, with a cumulative proportion of 10% reaching a plateau at years 5–10, which was similar to the 3.2%–9.7% rate reported over 5 years in a recent study adopting cyclophosphamide-ATG AHSCT.³¹ Rates of PIRA in our control group (40% at year 8) are overall similar to the roughly 5% per annum recently described in RR-MS by a systematic review,³² and lower than the 12% rate reported over median 1.6 years of treatment with NTZ or ocrelizumab in RR-MS.³³ In another study adopting a different cut-off for EDSS accrual, PIRA was reported in 24% of 184 RR-MS patients treated with NTZ for median 5 years.³⁴ However, comparisons between different studies should be made with caution due to heterogeneity in patient populations and definitions of PIRA.³²

AHSCT was superior to both NTZ and NTZ-oDMTs on relapses and NEDA-3, aligned with previous studies showing superior efficacy of AHSCT on relapse activity, NEDA-3 and time to disability progression compared with DMTs,³⁵ and on relapse activity compared with NTZ.²⁵ In the MIST trial, relapses and NEDA-3 failure at year 2 were reported in 75.4% and 74.7%, respectively, of control patients receiving NTZ, being both proportions remarkably worse than those observed in the AHSCT arm.³⁵

The cumulative proportion of converters to SP-MS was lower in the AHSCT compared with the CTRL group, the latter aligned with the 27% rate reported at mean 8 years of follow-up in the TYSTEN cohort,³⁰ and with the 27.6% rate at mean follow-up of 11.8 years in a recent Italian MS Registry Study,³⁶ both including patients with similar baseline characteristics. This could be regarded as a remarkable achievement as NTZ was reported to reduce by 50% the rate for conversion to SP-MS when compared with natural history.³⁷ In another study on AHSCT, rates of conversion to SP-MS were 18% and 4% after an average of 4 years in RR-MS patients treated with BEAM/ATG or cyclophosphamide/ATG, respectively, the former more disabled at baseline than the latter.³⁸ As MS phenotypes and transition to SP-MS were suggested to be determined by the interplay between consolidated tissue damage and exhaustion of individual CNS reserve,² the radical suppression of inflammation before tissue damage has achieved a critical threshold could effectively delay (or even abolish) conversion to SP-MS. Accordingly, age and EDSS at baseline independently predicted PIRA events, reinforcing the notion that timely treatment with HE-DMTs and AHSCT is crucial to control disease progression. In this respect, the recently published consensus statement from the European Committee for Treatment and Research in Multiple Sclerosis and the European Society for Blood and Marrow Transplantation endorsed the use of AHSCT as an appropriate escalation therapy for people with highly active MS failing HE-DMT—even a single HE-DMT in patients bearing markers of aggressive disease.¹⁰ When dealing with this latter category of patients, informed treatment decision-making can be supported by knowledge on long-term outcomes of HE-DMTs in real life, taking into account also the potential consequences of HE-DMT discontinuation and subsequent treatment switch.

The safety profile of AHSCT and DMTs was aligned with the literature. Two deaths due to advanced disease were reported in the CTRL group, and a similar signal for potential disproportion in long-term deaths was observed in the study by Kalincik et al (3/65 and 1/39 deaths in the NTZ and AHSCT groups, respectively)²⁷ and in a previous study from our group (2/62 and 0/31 deaths in cyclophosphamide and AHSCT-treated patients, respectively),³⁹ overall suggesting that risk for the worst outcome is not negligible in people with aggressive MS.

This study has several limitations. First of all, the sample size is small, although similar to other single-centre comparative studies on AHSCT. Furthermore, the monocentric design of the study provides robust accuracy of the information collected, which was generated by a limited pool of neurologists specialised in MS management and trained on standardised clinical practice of the centre. As previously acknowledged, it cannot be excluded that post-NTZ de-escalation in some patients increased the risk for long-term disability in CTRLs. On the other hand, 23% of patients in the AHSCT group had previously failed NTZ treatment; therefore, the effect of AHSCT observed can be considered the lower bound for the true effect of AHSCT, that is, as the smallest plausible value for AHSCT’s impact on the outcomes analysed, given the available data and assumptions. MRI markers associated with PIRA, such as paramagnetic rim lesions, could not be assessed as dedicated MRI protocols on 3T machines were not available yet during the index period. Likewise, anti-CD20 therapies were not licensed in Italy for the treatment of MS over that period, preventing us from including a control cohort of patients treated with such DMTs. Finally, the rate of long-term disability accrual observed in this study could be overestimated as few patients were exposed to early treatment with HE-DMTs; therefore, these results could be generalised in populations with similar baseline characteristics.

Conclusions

This study provides further evidence on the comparative effect of AHSCT and DMTs on long-term outcomes in aggressive RR-MS, focusing for the first time on PIRA. Besides its effect on new focal inflammation, AHSCT reduced the risk of long-term PIRA and conversion to SP-MS compared with DMTs, and PIRA was independently predicted by age and EDSS at baseline. Although exploratory due to the small sample size, these results suggest that early treatment with AHSCT or other DMTs targeting inflammation on both the peripheral and CNS compartments prevents long-term PIRA, raising the intriguing question as to whether these could radically change disease course and patients’ lives. Prognostic markers applicable to individual cases are eagerly needed to timely identify patients with aggressive MS who could benefit from early treatment with AHSCT.

Data availability statement

Data are available upon reasonable request. Aggregated de-identified participant data will be shared upon motivated written request to the corresponding author.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involved human participants and was approved by the Comitato Etico Area Vasta Centro, Regione Toscana (approval number 17696_oss). Participants gave informed consent to participate in the study before taking part.