Myelodysplastic syndrome and immunotherapy novel to next in-line treatments

ABSTRACT

Patients with Myelodysplastic syndromes (MDS) have few therapy options for sustainable responses in the frontline setting, and even less after hypomethylating agent (HMA) failure in relapsed and refractory setting. The only potential cure is an allogeneic hematopoietic stem cell transplant which is an unrealistic option for the majority of MDS patients. Immunotherapy with checkpoint inhibition, CAR-T cells, and vaccine therapy few have shown promise in a variety cancer and have now been tested in patients with MDS. Most trials have focused on AML patients and included small numbers of MDS patients. Until now, a dedicated review of immunotherapy outcomes in MDS patients has been lacking. Thus, herein we review outcomes of MDS patients after immunotherapies on a variety of clinical trials reported to date.

KEYWORDS:

• High risk MDS
• myeloid
• immunotherapy
• CAR T
• DLI
• checkpoint inhibitor
• T-cell exhaustion
• immune evasion
• relapsed and refractory MDS

Introduction

The myelodysplastic syndrome(s) (MDS) are a heterogenous group of neoplasms arising from myeloid progenitor cells¹ with consequential ineffective hematopoiesis, dysplasia and peripheral blood cytopenias.^2,3 The pathogenesis is complex involving epigenetic alterations, splicing changes, etc. that lead to maturation arrest. Hence, clinical outcomes across the MDS disease spectrum can be highly variable.

The International Prognostic Scoring System (IPSS) and the revised IPSS (IPSS-R) differentiates categories of MDS based on prognosis (estimated survival and time to leukemia progression).^4,5 There are four prognostic categories per IPSS-R, in descending order of estimated survival these are: Low, Intermediate-1, Intermediate-2 and High-risk MDS (HR-MDS).⁶

The only cure for MDS is with an allogeneic hematopoietic stem cell transplant (HSCT) which is the ultimate example of immunotherapy wherein a graft versus tumor (GVT) effect eliminates the MDS clone.^7,8 However, the toxicities of this procedure limit HSCT to only a subset of young and fit patients (~15%).⁹

Hypomethylating agents (HMA) such as azacitidine (AZA) or decitabine (DEC) are the recommended frontline line therapy for transplant ineligible intermediate, high or transfusion dependent MDS.¹⁰ Indeed, randomized phase III trials with AZA vs best supportive care or low dose cytarabine (ara-c) demonstrated a survival benefit in higher-risk (IPSS) MDS patients (21–24 months in AZA arm vs 11.5 months in BSC vs 15.3 months in ara-c).^11–13 DEC similarly has improved PFS and reduced AML transformation.^11,14

The prognosis after HMA failure is poor with a median OS of 5.6 months, and currently there are no standard therapeutic options for such patients.¹⁵ Elderly patients or those with comorbidities are significantly vulnerable, and there is a dire need for alternate salvage regimens for high risk MDS patients after HMA refractoriness.^16,17

Therapeutic manipulation of the immune system with newly introduced agents such as checkpoint inhibitors or adoptive T cell transfer has shown promise in several clinical trials for hematologic malignancies such as lymphomas but has yet to demonstrate efficacy in myeloid malignancies. Indeed, several approaches have been attempted in the context of acute myeloid leukemia (AML) with reported successes. However, MDS patients often make up only a small subset of those trials. Therefore, the purpose of this review, is to detail the clinical outcomes of exclusively MDS patients enrolled on immunotherapy clinical trials reported to date.

Immune checkpoint inhibitors

The natural function of immune checkpoints is to restore antigenic self-tolerance through multiple mechanisms such as modulating co-stimulation or direct immune cell inhibition.¹⁸ Checkpoint receptors are reliant on cognate ligands to facilitate interaction and these are variably expressed by a variety of cell-types.^19,20 For example, expression of the inhibitory ligand CD47 on myeloid leukemia cell acts as an anti-apoptotic “don’t eat me” molecule that also increases functioning of immune-suppressor T- regulatory cells (Treg) resulting in the death of antigen-specific effector T-cells (Teff).^19,20When inhibitory ligand expression is low or the checkpoint is pharmacologically blocked, tumor cell apoptosis can occur in addition to T cell receptor (TCR) mediated cytolysis induced by Teff cells. Indeed, this occurs naturally in normal responses to foreign antigens such as viral, intracellular bacterial pathogens and cancers.

Neoplastic cells of myeloid origin have taken advantage of these immune checkpoints with upregulation of inhibitory ligands such as PD-L1, B7-1, B7-2, CD47 etc. at baseline.¹⁹ MDS cells have demonstrated the capacity to harness some of these immunosuppressive effects to facilitate their survival, longevity, and proliferation. The latter is most marked in higher risk MDS (HR- MDS) and RAEB (refractory anemia with excess blasts).^19,20 Conversely, higher expression of immuno-inhibitory receptors of these ligands such as CTLA4, TIM3, PD-1, and ICOS were found on T-cells of patients with treatment refractory MDS patients compared with healthy donors.^18,19 In summary, sufficient evidence exists that patients with MDS have altered immune response to aberrant myeloid clones partly mediated by immune checkpoint modulation. Moreover, treatment with hypomethylating agents (HMA) has been shown to upregulate several of these immune checkpoints such as PD-L1, TIM-3 and CD47 on MDS cells. Therefore, it is conceivable that checkpoint inhibitors (CPI) that have been FDA approved for various solid tumors, lymphomas could be effective in MDS with or without concurrent HMA therapy yet reported outcomes to date have been unimpressive and colored by high rates of toxicities.²¹ CPIs have induced serious immune-related adverse events in MDS patients which can be particularly problematic if used soon before or after an HSCT. This next section will focus on safety and efficacy outcomes reported from CPI-based clinical trials. Only select CPI trials with preliminary or completed data exclusively enrolling MDS patients and/or data for MDS patients that are distinctly reported separately in multi-disease/AML trials are represented in (Table 1).

Table 1. Select MDS studies with immune check-point inhibitors

Immune Checkpoint	Checkpoint Inhibitor	NCI Registration ID	Phase	Author, Year Reported	IPSS Prognostic Risk Category	Median Age (years)	Primary Outcome	HMA
PD-1	Pembrolizumab	NCT01953692	Ib	Garcia-Manero, 2016	Int-1 (36%) Int-2 (32%) High (25%)	73	ORR	Failure
		NCT03094637	II	Chien, 2019	Int-1 or higher	Not Repor	Safety ORR	Failure
						ted	OS	Naïve
	Nivolumab	**NCT02530463	II	Garcia-Manero, 2018	Low (4%) int-1(40%) int-2 (37%) High (15%)	71	Safety	Failure
								Naïve
		NCT01822509	I/Ib	Davids, 2020	Any	Not Reported	Safety MTD AE	Post-Transplant Relapse
PD-1 + CTLA4	Nivolumab + Ipilimumab	**NCT02530463	II	Garcia-Manero, 2018	Int-1 (29%) Int-2 (43%) High (29%)	69	Safety	Failure
								Naïve
CTLA4	Ipilimumab	NCT01757639	I/Ib	Zaidan, 2017	Int-1 (45%) Int-2/high risk (55%)	67	MTD Safety AE	Failure
		**NCT02530463	II	Garcia-Manero, 201	Low (4%) int-	71	Safety	Failure
				8	1(40%) int-2 (37%) High (15%)			Naïve
PD-L1	Durvalumab	NCT02775903	II	Zeidan, 201	✦Int,	Not Re	ORR	Naïve
				9	high, very high	ported
	Atezolizumab	NCT02935361	I/II	O’Connell, 2018	✦Int, high, very high	Not Reported	Safety Tolerability	Failure
TIM-3	MBG453	NCT03066648	I/Ib	Borate, 2020	✦ high, very	71	Safety Tolera	Naïve
					high		bility MTD Efficacy
CD-47	Magrolimab	NCT03248479	Ib	Sallman,	✦ Int, high	73	ORR, AE	Naïve
	(5F9)			2020	, very high			Failure
Immune Checkpoint	Intervention	Line Of Therapy	Number Of MDS Patients (n=)	**IWG CR & PR %(n=)	*IWG ORR ϕ	Median DOR/PFS/EFS	Median OS /Survival Probability (%)	Status	▵Adverse Events ≧ Grade 3
PD-1	PEM	Relapsed & Refractory	28	4% (1 PR)	14-25%	Not Reported	6 mo Survival 49% IPSS high: 29% int-2: 22% int-1 89%	Completed	7%
	PEM + AZA	Relapsed & Refractory	20	10% (2CR)	30%	Not reported	5.9 mo	Recruiting	None Reporte
		Frontline	10	20% (2CR)	70%		12.9 mo		d
	NIVO	Relapsed & Refractory	15	0	13%	EFS 7 mo	8 mo 1 yr survival 25%	Arm Closed	40%
	NIVO + AZA	Frontline	20	* 50% (10)	75%	EFS 10 mo	12mo 1 yr survival 50%	Completed	“27%
	NIVO	Relapsed & Refractory	7	43% (3PR)	43%	^1 year PFS 23%	^1 year OS 56%	Active Not Recruiting	21% ≧
PD-1 + CTLA4	NIVO + IPI	Relapsed & Refractory	8	1 CR	29%	Not Reported	8.4 mo	Recruiting	43%
	NIVO + IPI + AZA	Frontline	6	100% (3 CR)	50%	Not Reported	Not Reached	On Hold
CTLA4	IPI	Relapsed & Refractory	29	0	7%	Not Reported	9.8 mo	Completed	24%
	IPI	Relapsed & Refractory	20	*15% (3)	35%	EFS 6 mo	8 mo 1 yr suvival 45%	Completed	“33%
	IPI + AZA	Frontline	21	*38% (8)	71%	EFS Not Reached	OS Not Reached 1 yr survival 68%		Not reported
PD-L1	Durvalumab + AZA	Frontline	42	3CR (7%)	62%	8.7 mo	11.6 mo	Active Not	None Rep
	AZA		42	4CR (10%)	47%	8.6 mo	16.7 mo	Recruiting	orted
	ATEZO + G-DEC	Relapsed & Refractory	9	1CR (11%)	33%	Not Reported	Not Reached	Recruiting	17 total AE gr3 or higher
TIM-3	MBG453 + DEC	Frontline	19	26% (5 CR)	58%	Not Reported	Not Reported	Recruiting	40-50%
	MBG453 + AZA		13	10% (1PR)	70%	Not Reported
CD-47	Magrolimab + AZA	Frontline	39	45% (CR 14, 1 PR)	91%	DOR Not reached	5.8 mo survival 100%	Recruiting	Not reported
	Magrolimab	Relapsed & Refractory	4	Not reported	Not Reported	Not Reported	Not Reported

Abbreviations: ATEZO = Atezolizumab, PEM = Pembrolizumab, NIVO = Nivolumab, IPI = Ipilimumab, DEC = Decitabine, AZA = Azacitidine, G-DEC = Guadecitabine. IPSS = International Prognostic Scoring System, IWG = International Working Group, MDS = Myelodysplastic Syndrome, MTD = Maximum Tolerated Dose, ORR = Overall Response Rate, CR = Complete Response, PR = Partial Response, mCR = Marrow CR, HI = Hematologic improvement, gr = grade, mo = Months, OS = Overall Survival, DOR = Duration of Response, PFS = Progression Free Survival, EFS = Event Free Survival,

** all data are from the same trial with separate MDS cohorts

✦IPSS-R (Revised International Scoring System)

≁Last Reported Update (Year)

**IWG MDS: International Working Group (IWG) 2006 response criteria in myelodysplasia.

ϕORR: defined as CR, mCR, PR, or HI based on IWG 2006 response criteria

^PFS and OS for all study patients, including MDS (n = 28)

▵Highest grade reported for Adverse Events for all study patients

“ based on 2016 abstract with 12/35 patients reported gr3/4 AE:

Programmed death ligand-1 & receptor (PD-L1/PD-1) inhibitors

PD-L1 (B7H1) is universally expressed on antigen presenting cells (APC) as an inhibitory ligand for PD-1 on Teff. PDL-1 expression on MDS cells was found in some patients to be increased at baseline and sometime greater than two times normal in those with HMA refractory disease, suggesting a possible mechanism for immune evasion that would benefit from inhibition of the PD-1/PD-L1 interaction.¹⁸

Pembrolizumab

The monoclonal antibody, Pembrolizumab, is a PD-1 blocking agent.²² In 2016, Garcia-Manero et al presented early phase I data on 28 relapsed/refractory MDS patients treated with PEM after failure of response to prior epigenetic therapy.²³ Enrolled patients were IPSS high 25%, intermediate-2 (int-2) 32% and intermediate-1 (int-1) in 36%.

At a median follow-up of 5.6 months ORR to PEM monotherapy was: 1 (3.5%) partial response (PR), 3 (11%) marrow complete response (mCR) and 14 (52%) stable disease (SD). The 6-month OS was 49% with dismal probability of survival at 29% and 22% for IPSS high and int-2 patients respectively. Int-1 patients had a 1-year OS of 89%. Overall, monotherapy with pembrolizumab was well tolerated (7% treatment related adverse events), the most notable treatment related events were 2 cases of tumor lysis syndrome in those with excess blasts. To build upon the efficacy demonstrated in this trial, the same group, Chien et al. combined PEM with AZA in a phase II trial of 35 high risk MDS patients. Patients were stratified as therapy naïve MDS or had previously failed HMA therapy.²⁴ The overall best response rate amongst the HMA refractory MDS cohort (n = 20) was 30%, including 2 CR, 2 mCR, and 2 hematologic improvement (HI). The HMA naive cohort (n = 10) had an impressive 70% best response rate (2 CR, 2 mCR, 2 mCR + HI and 1 with HI alone). The 13 responding patients had complex cytogenetics (n = 2), or mutations in ASXL1(n = 4) and TP53 (n = 3) frequently associated with poor outcomes in MDS.²⁵ However, overall survival estimates of combination PEM + AZA combination in the HMA refractory cohort remained at a lowly 5.9 and 12.9 months for HMA naïve patients at a follow-up of nearly 11 months, thus not much different from what would be expected with HMAs alone. The combination appeared safe and the most frequent AE’s of any grade were neutropenia, arthralgia and myalgia.

Durvalumab

Durvalumab is a monoclonal antibody targets the ligand PDL-1.²⁶ Zaiden et al²⁷ evaluated responses in 84 treatment naïve IPSS-R higher risk MDS patients in an open label phase II study with equal randomization to durvalumab + AZA vs. AZA alone. Notably, this is one of the few randomized trials comparing an HMA alone to a combination of HMA and immunotherapy as front-line therapy. The trial ended early due to higher-than-expected rates of disease progression, death and lack of response in the treatment cohort. Yet, the AZA only arm had twice the number of censored patients (n = 19) compared to the combination (n = 10) and a lower ORR (47%) compared with the combination (61%) that was not statistically significant. There were no differences in PFS or OS (see Table 1) but the combination treatment was well tolerated with 7 grade ≤ 2 immune-related adverse events (iRAEs).²⁷

Nivolumab

Nivolumab is a humanized monoclonal antibody that binds PD-1.²¹ Nivolumab has been tested as part of cytarabine-based induction regimens for AML and MDS-EB as preclinical models demonstrated upregulation of PD-1 ligands.^21,28 However, its use pre- HSCT may be limited due to immune-related AE’s (irAE’s) and the severity of GVHD post-HSCT.^29,30 Nivolumab has been evaluated as an agent to enhance “graft-versus-tumor” effect to treat post-HSCT relapse of MDS. In this phase I trial, of 28 patients with various hematologic malignancies which included 7 (25%) with HR-MDS, 3 of them attained a PR.³¹ However, induction and worsening of graft-versus-host disease (GVHD) after nivolumab remains a major safety concern in this setting. Indeed, there were two fatalities from GVHD and two more experienced serious dose limiting toxicities in the entire treated cohort. In summary, use of checkpoint inhibitors pre- or post-HSCT maybe have beneficial anti-MDS effects, but this should be weighed against GVHD in individual patients and should only be offered in the context of a clinical trial.

As part of a large multi-arm phase II trial, Garcia-Manero et. al reported preliminary data on 90 MDS patients with relapsed/refractory or therapy naïve disease divided into 6 treatment cohorts.

There were 76 patients that received checkpoint inhibitor monotherapy in 4 cohorts^32,33 vs. double checkpoint inhibitors (n = 14) in 2 cohorts.³⁴ All patients in treatment naïve cohorts received AZA in addition to immunotherapy.

We will discuss 2 cohorts treated that were treated with nivolumab monotherapy here (see section on CTLA4 inhibitors for the remaining 4 cohorts).

Thirty-five patients were treated with nivolumab with or without HMA in two cohorts.^32,33 The group of RR MDS patients (n = 15) received single agent nivolumab with 6 individuals having ≧ grade 3 AE, most frequently pulmonary infection (40%) and febrile neutropenia (33%).³² As it lacked clinical activity this arm closed prior to completed enrollment. The therapy naive cohort (n = 20) received nivolumab with the addition of AZA with 50% CR/CRp responses.³³ This suggests a synergistic effect with HMA induced PD-1/PD-L1 upregulation with simultaneous PD-1 blockade on Teff cells with nivolumab.¹⁸

Atezolizumab

The PD-L1 inhibitor, atezolizumab,³⁵ was evaluated in a phase Ib trial with IPSS-R higher risk MDS who were older (median age: 76; range: 63–89) and stratified by whether they had failed prior HMA (where they received only atezolizumab or in combination with an HMA) or were treatment naïve (where they received combination HMA and atezolizumab).³⁶ This trial much like the durvalumab trial was halted early because of a high incidence of grade ≥ 3 adverse events (71%) in both arms of the study (71% and 57%, respectively) as well as excessive early deaths (29%) within the naïve cohort compared with historical controls. In contrast, atezolizumab in combination with Guadecitabine, a next-generation HMA, was well tolerated in 9 relapsed and refractory MDS patients.³⁷ There were no DLT’s after a median of 5 treatment cycles and 94% of grade 3/4 AE’s were related to peripheral cytopenias.

Cytotoxic T-lymphocyte-associated protein 4 (CTLA4) inhibition

Ipilimumab

The T cell expressed surface protein CTLA4, dampens T cell response after binding its ligands: CD80 (B7-1) and CD86 (B7-2) which are typically expressed by antigen presenting cells. Ipilimumab (IPI) effectively blocks CTLA4 function thereby activating Teff resulting in clonal proliferation of antigen-specific T cells.³⁸ Simultaneously, IPI has been shown to downregulate immunosuppressive functioning of Treg cells through its action within lymph nodes and within the cancer micro-environment.³⁹

Much like PD-L1 expression, MDS cells from untreated patients have been shown to overexpress CTLA4 which can be enhanced by HMA therapy.¹⁸

A multicenter phase I trial of IPI established a maximum tolerated dose of 3 mg/kg at which no grade ≥ 2 immune-related adverse events were seen in 29 HMA-refractory MDS patients.⁴⁰ In this refractory cohort, IPI as a single agent induced a 21% clinical benefit rate (defined as stable disease for over 12 months) of which 2 (9%) were mCRs. Furthermore, no early excessive deaths were observed (death rate of 15%) and in five who went onto get an HSCT, there was no increased incidence of GVHD post-HSCT either. Responding patients had significantly higher expression of inducible T cell costimulatory molecule (ICOS) on CD4+ (p = .05) and CD8+ (p = .01) T cells in comparison with those who failed therapy.⁴¹

Garcia-Manero reported phase II safety and preliminary responses for 41 MDS patients treated with ipilimumab in 2 cohorts.^32,33 Relapsed and refractory patients received single agent IPI with manageable side effects³² and ORR of 35% (n = 7).³³ Therapy naïve patients received IPI + AZA and had ORR 71% with 38% of responders having CR/CRp.³³

In conclusion, the 4 treatment cohorts (including those in the section on PD-1 inhibitors) that received checkpoint inhibitors with or without AZA had a favorable safety profile and tolerable toxicities, with rash and fatigue being the most common AE of any grade. Responses in the combination groups with HMAs were encouraging and perhaps higher than would be expected with HMAs alone, yet confirmatory trials are pending.

CTLA4/PD-1 combined inhibition: ipilimumab with nivolumab

In the Garcia-Manero multi-cohort phase II trial, 8 patients were treated on an arm that received a combination of ipilimumab and nivolumab to treat RR MDS while 6 others were treated on another arm that included treatment naïve patients who received combination CPIs with an HMA. The RR group of 8 MDS patients had an ORR of 29% with 1 CR and 1 HI while 3 of 6 in the treatment naïve cohort achieved CRs.³⁴

Taken together, checkpoint inhibitors of CTLA4 and PD-1 pathways that have been highly effective in select lymphomas and refractory solid malignancies, have yet to demonstrate similar results in patients with MDS. Combination approaches with HMAs have been encouraging but direct comparisons with HMA therapy alone is still in progress. The reasons for minimal single-agent efficacy in MDS patients might be multifold including lack of immunogenicity of MDS cells, defective T cell function in patients with MDS etc, all of which are under preclinical investigation.

T cell immunoglobulin and mucin domain-3 (TIM-3) inhibition

The immune checkpoint receptor T-cell Immunoglobulin and mucin domain 3 (TIM-3) is expressed on monocytes, dendritic cells, and Teff. Interaction with its ligand, galectin-9, leads to suppression of T cell activation. Like all previously described checkpoints, Tim-3 is expressed at a higher level on exhausted T cells.⁴² The ligand is overexpressed preferentially on leukemic stem cells and myeloid blasts compared to normal hematopoietic stem cells and has been recognized as a potential target for AML/MDS immunotherapy.⁴²

MBG453

MBG453 is an anti-TIM-3 monoclonal antibody that has been tested in the clinic for its dual capacity to induce antibody-dependent cytotoxicity as well as its ability to enhance T cell activation.

Borate et al.^43,44 reported separate outcomes for therapy naive MDS patients (Table 1) that were included in a phase I treatment arm with MBG453 + HMA using AZA or DEC. The MBG453 + DEC cohort has treated 19 HR-MDS in its most recent report, in whom at ~5 months post-treatment, 58% (11/19) of patients achieved a clinical response including 5 CRs. Ten of the 13 patients that received MBG453 + AZA were evaluable for responses with 1PR and 6 mCR.

Longer follow-up and comparison with standard of care cohorts will determine if this response rate while promising is different from HMAs alone. Immune-related adverse events with MBG453 were reassuringly low and only 1 patient experienced elevated liver enzymes (at grade 3). MBG453 added to HMA appeared to have a favorable safety profile, with response durations that might be longer than would be expected for DEC or AZA alone, however, the full report is eagerly awaited.^43,44

CD-47 inhibition

The CD47 ligand SIRP-α is universally expressed on human cells and serves as a “don’t eat me” signal for the innate immune system, specifically macrophages.^45,46 Reproducibly, MDS cells have been shown to greatly overexpress CD47 possibly to evade macrophage mediated killing while enhancing antiapoptotic and proliferative signals.^39,47 Indeed, its overexpression correlates with higher risk MDS as well as poor overall survival.^47,48 So, strategies to block the CD47-SIRP-α interaction may enhance macrophage mediated phagocytosis and have now been exploited in the clinic with potential agents in clinical development discussed below.^47,49,50

CC-90002

CC-90002 is an anti-CD47 humanized IgG4 monoclonal antibody, which was tested in a phase 1 trial of relapsed or refractory high risk MDS (n = 4) and AML (n = 24).⁵¹ Unfortunately, treatment resulted in excessive serious treatment related AE’s (SAE’s) in 82% of all dose escalation study patients (0.1 mg/kg up to 4 mg/kg) CC-90002 and 16 on-study deaths. Four patients experienced a serious DLT with grade 4 disseminated intravascular coagulation and grade 4 cerebral hemorrhage, grade 3 purpura, grade 4 congestive cardiac failure and grade 4 acute respiratory failure, and grade 4 sepsis. Perhaps the advanced blast burden or the dosing schema as well as the immunogenicity of the monoclonal antibody used may have contributed to the observed hyperinflammatory/autoimmune process.^52–54 The drug had poor preliminary efficacy in all study patients with best responses in those with MDS patients having SD (n = 2), causing the sponsor to terminate the trial.

Magrolimab (Hu5F9-G4 or 5F9)

By contrast, the humanized and thus probably less immunogenic anti-CD47 monoclonal antibody, Magrolimab (aka Hu5F9-G4 or 5F9) did not demonstrate similar toxicities^55,56 in a phase Ib dose escalation study of IPSS-R Int-2 or high risk MDS. All patients received a priming dose of 5F9 (1–30 mg/kg weekly) that was done to mitigate hemolysis⁵² which was reported in a prior study with an unclear mechanism of action.⁵³ This “prime” may have also potentially offset severe AE’s seen with CC-90002. MDS patients enrolled were either treatment naïve (n = 39)⁵⁷ or relapsed/refractory (n = 4).^55,56 In combination with HMA, the investigators reported an impressive ORR of 100% with 54% CR and 39% mCR + HI. Two of these patients had a baseline tp53 mutations and both achieved an MRD negative CR. The responses were rapid with a median 1.9 months to response which is much faster than expected for AZA alone (up to 4 cycles of 4 months).⁵⁶ Indeed most of these responses lasted for longer than 4 months, as reported from an update presented in 2020, where ORR at ~6 months remains high at 91% (30 of 33 still in response) with no deaths reported (100% OS). Those who had failed HMA, received monotherapy with 5F9, where no responses were seen.⁵⁵ However, both arms tolerated treatment with 5F9 well (MTD not reached) and the most common adverse events were cytopenias (<25% developed any cytopenia).

Based on the results of the MDS cohort a phase III National Cancer Institute (NCI) clinical trial has started recruiting 9/2020 registered at NCT04313881 as a double-blind placebo controlled comparing magrolimab in combination with AZA vs. the standard of care AZA + placebo in therapy naïve higher risk MDS patients. Results from this and other ongoing efficacy-based trials for this combination are eagerly awaited.CPI have demonstrated objective responses in early phase trials with higher-risk MDS patients however not to the extent observed with lymphomas or select solid tumors. With the caveat of cross trial comparisons, the CR+PR and ORR of patients treated with CPI + HMA is safe and at least similar to standard of care therapy in the frontline setting with some indications of additive benefit.^11,14 Unfortunately, MDS patients that have previously failed HMA therapy, remain an unmet need and CPIs like many other treatments for this cohort have been largely unsuccessful at inducing high response rates.

There is a trade-off with increasing severity and incidence of adverse events particularly with utilization of two CPI drugs. Most irAE’s were reversible with steroid administration and drug holding with safe resumption of the CPI. However, in the real world this may limit its widespread use in elderly patients with poor PS and other comorbidities.

As with other malignancies, the search for biomarkers that may predict response to check-point inhibitors for MDS/AML is ongoing and might be necessary before accepting CPIs as a bonafide treatment modality for MDS.

Adoptive T-cell transfer therapy

In malignancy, adoptive cellular therapy is the direct killing of tumor cells by specific effector lymphocytes. In its simplest form, adoptive T-cell transfer is the infusion of T-lymphocytes into a patient’s body to directly target an antigen, usually for cytotoxicity. It can be autologous or allogeneic T-cell transfer, which can be unmanipulated (as in simple donor lymphocyte infusion) or manipulated such as CAR (Chimeric Antigen Receptor) T-cell constructs, modified T-Cell receptors, or ex-vivo antigen exposure priming with or without genetic modification.

The myeloid malignancies have not had the same success to date as their lymphocytic counterparts, as there are many shared antigens with progenitor stem cells. However, we are finding unique ways to manipulate lymphocytes and target myeloid antigens in a more streamlined fashion without the toxicities seen in other immunotherapies to date, although this may be related to their lack of engraftment and/or expansion in vivo.

Here we will discuss select therapies with CAR-T and manipulated DLI for patients with MDS.

Chimeric antigen receptor

Autologous lymphocytes can be engineered to recognize and target antigens completely independent of MHC or HLA restriction. In other words, they are continuously programmed to bind and recognize the target without needing TCR nor co-stimulation in second generation formats, as activating receptors (such as CD28) can already be fused to the intracellular CD3 TCR complex.⁵⁸

Natural killer group 2 (NKG2) receptors

The natural killer group 2 receptors (NKG2D)⁵⁹ are a group of C-type lectin immunomodulatory proteins found on natural killer (NK) and CD8 + T cells. In response to intracellular stressors such as DNA damage, toxins, infection, inflammation and cancer, its ligands: MHC I chain-related proteins (MIC) MIC-A, MIC-B molecules and the activating unique long 16-binding proteins (ULBP 1–6) are upregulated on affected cells (eg. cancer cells).⁶⁰

Upon activation there is simultaneous rapid NK-mediated cytotoxicity as well adaptive T cell immune response that leads to marked increases in circulating levels of IFNγ and TNFα.⁶¹ The same ligands that activate NKG2D receptors are cleaved into soluble forms forming of negative feedback to downregulate receptor expression and dampen the inflammatory process.⁶⁰

As NKG2D ligands are only expressed under pathologic inflammatory and not in the normal state, it makes an attractive target for CAR-T cell therapy. However, its expression is dynamic and fluctuates considerably. Nevertheless, T cells expressing CARs that mimic NKG2D receptors have been clinically tested with reported successes.

NKG2D-CAR T cells

T cells genetically engineered to express NKG2D have demonstrated the potential to target and specifically kill not just cancer cells but also modulate Tregs and myeloid-derived suppressor cells within the tumor microenvironment.⁶² Furthermore, in these preclinical models, genetic overexpression of the receptor overcomes the natural inhibition mediated by cleaved soluble ligands like MIC-A.^61,63,64 NKG2D-CAR T cells rely on DNAX-activating protein of 10kDa (DAP10) endogenous expression to stabilize the NKG2D receptor which is in contrast with CD28 or 4–1BB co-stimulation required for CD19 CAR T cells.^58,65–67

Clinical trials of NKG2D CAR-T cells have included patients with MDS because MDS cells, much like AML or myeloma cells have also been shown to greatly overexpress the ligands for NKG2D.^62,68 We will review the available data for two investigational NKG2D CAR-T therapies that include MDS patients on protocol. Only select Adoptive Cellular therapy trials with preliminary/completed data exclusively enrolling MDS patients and/or data for MDS patients that are distinctly reported separately in multi-disease/AML trials are represented in Table 2.

Table 2. Select MDS studies with adoptive cellular therapies

	Native T-cell Receptor	Genetically Modified Construct	Antigen Specific Target	CAR-T Product	Phase NCT	Author, Year. Reported	Lymphodepleting CT Regimen**	Total No. Patients (n=)⍦	Line Of Therapy
CHIMERIC ANTIGEN RECEPTORS	NKG2D	NKG2D + CD3ζ		CYAD-01	I (THINK) NCT03248479	Sallman. 2019	No	25	Relapsed & Refractory
					I/II (DEPLETHINK) NCT03466320	Al-Homsi 2019	yes	9	Relapsed & Refractory
				CYAD-02	I (CYCLE-01) NCT04167696	TBD: ASH, 12/2020	yes	NA	Relapsed & Refractory
				CM-CS1		Baumeister 2019	No	11	Relapsed & Refractory
					NCT04167696
		Multi Leukemia Antigen Specific Targets		T- Cell Source	Phase NCT	Author, Year Reported⏆	Preconditioning Chemotherapy	Total No. Patients (n=)⍦	Line Of Therapy
DONOR LYMPHOCYTE INFUSION		WT1, PRAME, Survivin, NYESO1		Donor DERIVED	I (ADSPAM) NCT02494167	Lulla 2019	Yes	21	Post-HSCT adjuvant (n=12)
									Post-HSCT- Relapsed & Refractory (7)
Primary Outcome	Intervention (CAR-T cell Schedule):	INDUCTION Dosing Level Cohorts (T-cells/inf)	Number of treated MDS Patients (n=)	✦IPSS-R / WHO 2016 Classification	Hypomethylating Agent ℥	*ϕMDS Responses	Status	* ≧ Grade 3 AE	≧ Grade 3 CRS
DLT	CAR-T cell Every 2-weeks x 3	Two week intervals: DL1: 3x108 DL2: 1x109  DL3: 3x109	3	Intermediate (n=1)	Failure	1mCR	Recruiting	44%	24% (n=6)
	CAR-T cell dosed Every 1-week x 4	Dose Dense: DL2:1x109  DL3: 3x109		High/Very High (n=2)		0
Safety	CYFLU--> CAR-T cell dose x 1	DL1: 1x108	2	Very High (n=2)	Failure	not reported	Active. Not Recruiting	** 22%	*n=2 (22%)
	CYFLU--> CAR-T cell dose x 1 on Day 3	◦ DL2: 3x108	NA	NA	NA	NA	NA	None Reported
Safety	CYFLU--> CAR-T cell dose x 1 on Day 3	DL1 1x108 DL2: 3x108  DL3: 1x109	NA	Intermediate, High Very High	Failure	NA	Recruiting	TBD	TBD
MTD	CM-CS1 dose escalation	3×107 = MTD	6 MDS/AM	Higher Risk (RAEB)	Failure	0	Completed	0	0
	cohorts		L
Primary Outcome	Intervention	Dose Levels T-cells/m2 per infusion	Number of treated MDS Patients (n=)	✦IPSS-R / WHO 2016 Classification	Hypomethylating Agent ℥	*ϕMDS Responses	Status	▵ ≧ Grade 3 AE
DLT	HSCT-----> TAA--DLI day +30	DL1: 5 x 10e6 DL2: 1 x 10e7 DL 3: 2 x 10e7 DL4: 5 x 107 DL5 1 x 108	2	NA	NA	2 CR	Recruiting	0

Abbreviations: CAR = chimeric antigen receptor, cCAR = compound CAR CAR-T Chimeric Antigebn Receptor T-Cell,DL = Dosing Level, DLI = Donor Lymphocyte Infusion HSCT = Hematopoetic Stem Cell Transplant; NKG2D = Natural Killer Group 2D-R receptor, NKG2DL – Natural Killer Group 2D Ligand; IPSS = International Prognostic Scoring System, IWG = International Working Group, MDS = Myelodysplastic Syndrome, RAEB = Refractory Anemia Excess MTD = Maximum Tolerated Dose, ORR = Objective/Overall Response Rate, CR = Complete Response, PR = Partial Response, mCR = Marrow CR, HI = Hematologic improvement, gr = grade, mo = Months, OS = Overall Survival, DOR = Duration of Response, PFS = Progression Free Survival, CRS = Cytokine Release Syndrome, TBD = To be determined

* *Preconditioning Regimen Prior to CAR-T administration: CyFlu: Cyclophosphamide 300 mg/m²/day and fludarabine 30 mg/m²/day, daily for 3 days.

✦IPSS-R (Revised International Scoring System) Last Reported Update (Year)

℥ HMA Response prior to enrollment

⍦ Includes all patients on study with MDS and non-MDS hematologic cancer

*IWG MDS: International Working Group (IWG) 2006 response criteria in myelodysplasia.

ϕ ORR: defined as CR, PR, mCR or HI based on IWG 2006 response criteria (DL3: 1 × 109 no data yet)

◦No data on DL3 at the time of the report

℥ HMA response prior to enrollment

▵Highest reported Adverse Events of All Hematologic Cancers Enrolled and Treated (includes non-MDS)

CM-CS1

Baumeister et. al. published the first phase I trial of a NKG2D CAR T cell (called CM-CS1) with single dose administration of fresh product in 4 escalation cohorts with R/R MDS/AML (n = 6) and MM (n = 5) patients.^67,69 The highest infused dose was 3 × 10⁷ T cells. Compared with other autologous CAR-T cell products, CM-CS1 was safe and well tolerated, administered without lymphodepletion. Levels of CM-CS1 CAR-T cells were undetectable after 1–2 weeks in most patients. Unfortunately, there were no responses per disease specific response criteria at day 28 post infusion, and 9 evaluable patients had PD.⁶⁷

CYAD-01

Celyad pharmaceuticals manufactured the NKG2D-CAR (NKR-2) T-cell product, named CYAD, for use in the multi-national phase I THINK (THerapeutic Immunotherapy with NKR-2) trial with various cancer types.⁶¹ In contrast to CM-CS1, CYAD-01 requires cryopreservation for multiple NKR-2 injections from single patient apheresis.⁷⁰

Sallman et al.⁷¹ presented outcomes in 25 relapsed and refractory patients (3 MM, 19 AML, 3 MDS) receiving CYAD-01 CAR-T cells without prior lymphodepleting CT (THINK trial). All 3 MDS patients (ages 80–85) had HMA refractory disease. The most common grade ≥3 adverse event was CRS which occurred in 24% of all treated subjects (6 of 25) and resolved completely post-tocilizumab. One out of 2 MDS who were available for response assessment had a transient morphological CR after three infusions of CYAD-01 at the highest DL3 (3 x 10⁹ cells). However, after week 12 despite a second infusion cycle the patient experienced disease progression. Overall, in the entire cohort, the duration of response in all 5 responding patients was brief (~4-8 weeks in the majority). There was no correlation between the dose levels of CAR-T cells infused on obtaining a response.

Of note, the THINK trial was conducted without the administration of lymphodepleting chemotherapy. Preclinical models have since demonstrated that cytotoxic chemotherapy upregulates NKG2D ligand expression, decreases Treg activity and thus enhances NKG2D CAR-T cell expansion.^62,68,72,73 This led to the DEPLETHINK phase 1 trial⁷⁴ for patients with relapsed/refractory AML or MDS where patients were treated with lymphodepleting conditioning with cyclophosphamide and fludarabine prior to a single dose of CYAD-01 CAR-T cells. When comparing expansion metrics of the THINK and DEPLETHINK participants, the expansion and perhaps even persistence was superior in those that received lymphodepleting chemotherapy, however, this did not translate into clinical response as 2 patients with MDS on DEPLETHINK had no objective response by week 4 post-infusion. The higher CAR-T cell expansion resulted in higher frequencies of serious toxicities that were dose dependent. Indeed, 2 out 3 patients that received chemotherapy experienced grade 3 or higher CRS and both experienced encephalopathy syndromes consistent with immune effector cell-associated neurotoxicity. Novel strategies during manufacture of NKG2D CAR-T cells as well as new constructs (CYAD-02) have demonstrated superior efficacy to CYAD-01 in preclinical models because they address some major barriers such as fratricide or target specificity of NKG2D-CAR T cells are currently under investigation. Results from clinical trials of these new NKG2D CAR-T cell products are eagerly awaited.

Donor directed lymphocytes

Relapse after allotransplant for higher risk MDS patients has a dismal outlook with poor survival.⁷⁵ Consideration of unmanipulated donor cell infusion (DLI)⁷⁶ or second allo-HSCT is dependent on the duration of the previous remission period.^77,78 However, neither of these options have durable responses and DLI is unpredictable with life-threatening GVHD.

Patient tailored DLI with tumor-associated antigen (TAA) stimulation ex vivo selects for an enriched, polyclonal CD4+ and CD8+ specifically directed against myeloid malignancies, such as AML and MDS. Our group has been able to mitigate GVHD with non-manipulated DLI by selecting T cells from DLIs that are specific for AML/MDS antigens typically not present on normal host cells, such as WT1, PRAME, Survivin and NYESO1.⁷⁹

Data presented thus far have shown this approach to be safe in 19 patients with high-risk AML/MDS (of which 2 had MDS). These patients had either relapsed after HSCT or had an unacceptably high risk of relapse due to the nature of their underlying disease. The latter group of patients that received TAA-T cell prophylaxis (n = 12) had 4 relapses <1 year from the infusion, but 2 were subsequently re-treated and one went onto another line of therapy. In summary, 11 of 12 patients that received the donor-derived TAA-T cells remain alive and in CR.

The responders had measurable expansion of leukemia antigen-specific T-cells that persisted for >9 months and exhibited durable memory. Impressively, there was no GVHD related to the infusions which was likely offset by selecting for “leukemic-specific” targets that are not present on normal cells. Confirmatory comparator trials are now underway (NCT04511130).

The use of preprogramed direct cytotoxic T-cells makes adoptive cellular therapy with autologous CAR T-cells and TAA- DLI attractive options for RR MDS.

Donor derived killer T cells are primed ex vivo with a selective immune signature tailored to the patients’ MDS cells. Thus far, TAA-DLI rapidly achieves disease control and appears to be safer, with respect to GVHD, than unselected donor T-cell infusion in the post-transplant setting. Of course, this may not be produced on a large scale and is highly dependent on having donor availability.

The utilization of autologous CAR T cell therapy in myeloid malignancies and particularly in MDS remains uncertain. There are many obstacles that must be overcome to be a successful therapy option for RR high-risk MDS, and their inability to persist with memory is problematic. Certainly, the CAR T cell expansion process is improved with lymphodepleting chemotherapy prior to infusion. This further reinforces the central dogma that within the bone marrow microenvironment of MDS there is profound immunosuppressive effects and impaired immune surveillance. Key players such as myeloid-derived suppressor and regulatory T-cells must be intervened upon for to have any meaningful therapeutic effect.^80,81

On the horizon are constructs with multiple MDS antigen targets such as compound CAR T’s cells (cCAR) and Natural killer CAR’s, and NK-cell based CAR therapies.

Therapeutic vaccines

For more than 2 decades, various myeloid leukemia antigens have been identified as targets for vaccine-based T cell immunotherapies. The intention of preclinically and clinically tested vaccines is to stimulate a “tumor-specific” pool of T cells with specificity mediated through their native T cell receptor (TCR). Strategies have also focused on recruiting CD4+ helper T-cells by presenting peptides in the context of major histocompatibility class (MHC) II molecules by professional antigen presenting cells (APCs). This would not only give a robust delayed type hypersensitivity reaction with enduring memory, but it could also reactivate the exhausted T-cells with immunologic dormancy. Here-in we will discuss therapeutic MDS vaccine strategies with various platforms including peptide tumor associated antigens (TAA), dendritic cell based and autologous tumor cells. Only select vaccine trials with preliminary/completed data exclusively enrolling MDS patients and/or data for MDS patients that are distinctly reported separately in multi-disease/AML trials are represented in Table 3.

Table 3. Select MDS studies with vaccines

	Vaccine	Antigen Specific Target	HLA Epitopes	Adjuvant/Delivery	NCI Registration ID	Phase	Author, Year Reported	Line Of Therapy	Primary Outcome
WT1	WT4869	WT1	HLA-A*24;02	Not Reported	Japanese trial	I	Suzuki, 2015	Relapsed and Refractory	Safety
	DSP-7888	WT1	HLA-A*02:01/06 HLA-A*24:02 HLA-DRβ1	Not Reported	NCT02436252	I/II	Miyakoshi, 2016	Relapsed and Refractory	Safety & OS
	WT-1 peptide vaccine	WT1-A1 WT1-122A1long WT1-427long WT1-331long	HLA‐A*02:01 HLA-DR.B1	Montanide ISA51VG + GM-GCSF	NCT00665002	I	Brayer, 2015	Relapsed and Refractory	Safety Tolerance Immunogenicity
PR-1	proteinase 3. PR1 peptide	PR-1	HLA-A2	Montanide ISA51VG + GM-GCSF	NCT00004918	I/II	Qazilbash. 2017	Relapsed and Refractory	Toxicity and immunogenicity
PR1 + WT1	PR1	PR1:169-177 WT1:126-134	HLA-A*0201	Montanide ISA51VG + GM-GCSF	NCT00270452	I	Rezvani, 2008	Treatment Naïve	Safety Toxicity
	PR-1	PR1:169-177 WT1:126-134	HLA-A*0201		NCT00499772	II	Rezvani, 2011	Relapsed	Immunogeniity
NY-ESO-1	DEC205 mAb + NY-ESO-1 fusion protein (CDX-1401)	NY-ESO-1	HLA-unrestricted	poly-ICLC adjuvant	NCT01834248	I	Griffiths, 2018	Treatment naïve	Toxicity
	Vaccine Dosing Interventions:	MDS Patients (n=)	✦IPSS-R/IPSS risk/ or WHO Classification^ (n=)	Median Age (years)	Hypomethylating Agent ℥	*IWG CR & PR	*IWG ORR ϕ	Median OS /Survival Probability (%)	Status	* ≧ Grade 3 AE
WT1	5 up to 1,200 µg/body q 2 weeks	26	HR MDS (17) LR-MDS (9)	Not Reported	Failure	0	18.2%,	13 mo**	Completed	0
	3.5-10.5 mg/body,Id every 2-4 weeks	12	HR-MDS (7) LR-MDS (5)⌑	Not Reported	Failure	0	17%	7.3-10.8 mo**	Active Not Recruiting	□ 50%.
	Vaccine q 2 weeks x 6	2	Int2/High risk	74	Failure	0	0	Not Reported	Completed	6%
PR-1	Vaccine Dose every 3 weeks x 6	11	RAEB I, RAEB II, RA, RA	69	Failure	1 PR	36%	Not Reported	Completed	0
PR1 + WT1	single dose	2	RA & RARS	41	Naïve	0	0	Not Reached	Completed	0
	Vaccine x 2 weeks x 6	2	RA & RARS	41	Naïve	0	0	Not Reached	Completed	0
NY-ESO-1	Vaccine x1 alone cycle1, Vaccine C2day 15 + Decitabine d 1-5 x 4 cycles	7	Very High (4) High (1) Intermediate(2)	65 yo (56-79)	Naïve	3	71%	Not Reported	Completed	⍜ 66%

Abbreviations: CHLA = Human Leukocyte Antigen; IPSS = International Prognostic Scoring System, IWG = International Working Group, MDS = Myelodysplastic Syndrome, RAEB = Refractory Anemia Excess MTD = Maximum Tolerated Dose, ORR = Objective/Overall Response Rate, CR = Complete Response, PR = Partial Response, mCR = Marrow CR, HI = Hematologic improvement, gr = grade, mo = Months, OS = Overall Survival, DOR = Duration of Response, PFS = Progression Free Survival

≁ Last Reported Update (Year)

✦IPSS-R (Revised International Scoring System)

^WHO 2016 classification: RA = Refractory Anemia, RAEB = Refractory Anemia with Excess Blasts I and II, RARS = Refractory Anemia with Ringed Sideroblasts

*IWG MDS: International Working Group (IWG) 2006 response criteria in myelodysplasia

ϕORR: defined as CR, PR, mCR or HI based on IWG 2006 response criteria

□lower risk MDS but transfusion dependent

℥ prior HMA use and/or response in HR-MDS

℥HMA response prior to enrollment

**med OS i higher risk patients

▵Highest reported Adverse Events of All Hematologic Cancers Enrolled and Treated (includes non-MDS)

□ (most injection site reactions) ⍜Most were hematologic toxicities

Peptide vaccines

Wilms’ tumor-1 (WT1)

Wilm’s tumor (WT1) gene encodes various regulatory functions involved in cellular differentiation and proliferation.⁸² It has oncogenic potential with overexpression and association with bcl-2 and p53 genes, amongst others. In patients with MDS, there is a strict direct correlation between high WT1 mRNA levels and marrow blast percentages.^83,84 Furthermore, higher levels of WT1 in peripheral blood correlate with poor response to HMA therapy and thus an inferior OS.⁸⁵ WT1 is a suitable immunotherapeutic target, as WT1 expression is significantly higher in MDS patients compared with healthy controls and is relatively immunogenic to T cells.^83,86–88 Pilot studies of peptide vaccines of WT1 to date have demonstrated the capacity to induce clonal expansion of WT1 reactive T cells (predominantly CD8 + T cells).^89–91 Indeed, >50% of all vaccine recipients across the tested peptide vaccines develop WT1-peptide reactive T cell expansions. In all the trials reported to date the expanding WT1 reactive T cells in response to MHC I or II bound peptides did not induce any significant adverse events or autoreactivity (Table 3). For example, Suzuki et al.⁹² demonstrated no grade ≥3 events at the highest tested therapeutic dose of their peptide vaccine and similarly, the vaccine DSP-7888 which contains both Class I and class II peptides did not demonstrate any grade ≥3 toxicities at the highest tested doses.⁹² There were 2 SAE’s with myocarditis and fever, unrelated to the dose. The predominant response-type reported among MDS patients to date has been disease stability. There have been select cases of exceptional responders with Suzuki et al. reporting 4 of 22 obtaining an objective response (1 mCR and 3 HIs). Brayer et al.⁹¹ reported an elderly MDS patient who was HMA refractory but achieved long lasting (for 22 months) transfusion independence post vaccination and Keilholz et al.⁹³ reported 1 MDS patient out of 2 with complete recovery of neutrophil counts post-vaccination that persisted for 10 months. In all trials, authors reported longer than expected disease stability and survival which might reflect an alteration in the natural course of the disease induced by T cell pressure. So future strategies that can expand the breadth of target peptides (as a combination with targeting other TAAs, see below), combinatorial approaches with complementary agents such as HMAs (to enhance expression of WT1) or checkpoint inhibitors might be necessary to induce deeper and longer lasting clinical responses.

Proteinase-3+ neutrophil elastase (PR-1)

Proteinase 3 (PR3) is present in high concentrations within the cytosol of myeloid leukemic blasts. By contrast normal granulocytes express PR3 only within the granules with minimal processing in the cytosol. Indeed, PR3 peptides have been shown to be presented naturally by only leukemic blasts and not normal granulocytes in the context of HLA-A2.⁹⁴ Endogenous responses to PR-3 have been shown to exist in HLA-A2+ patients with AML or MDS but it has been hypothesized that the chronic overexpression may lead to immune tolerance and exhaustion with selective clonal apoptosis of PR3-specific memory T cells.⁹⁵ Thus PR3 is considered an attractive target for vaccine immunotherapy of MDS. The largest trial of a PR3 vaccine was reported by Qazilbash et al.⁹⁶ where in 4 of 11 patients with MDS who received HLA-A2 restricted peptide-based PR3 in combination with neutrophil elastase, called the PR-1 vaccine⁹⁶ had a clinical response (1 PR and 3 HIs). As with tumor-based vaccine trials, the vaccine was safe at all tested doses despite demonstrating >2-fold increase in peripheral PR3-specific CD8+ specific T cells in 2 of the 4 responding patients. Similar to the WT1 vaccine approaches, this trial demonstrated that PR3 vaccines can be used in older MDS patients unfit for aggressive therapies as well as a safe immunotherapeutic approach in the post-HCT settings.

PR1 and WT1 combination vaccine

In an attempt to broaden the immune response to MDS clones, Rezvani et al.⁹⁰ administered a HLA-A2 restricted PR1 as well as a WT1 peptide vaccine to 2 patients with MDS. While neither patient had an objective response, both mounted PR1- and WT1-specific T cells post-vaccine that correlated in a decline in mRNA transcripts of WT1 and PR1 and this translated into long-term disease stability lasting for >2 years in both cases.

NY-ESO-1

NY-ESO-1 is one of several cancer testes antigens (CTA) expressed in trophoblastic tissue, spermatocytes and germline cells, but not in normal post-meiotic cells.^97–100 Germline cells naturally lack MHC I expression to promote an immunoprotective environment against potential self-destruction. Therefore, CD8+ T-cells cannot recognize CTA expression on these immune privileged cells.⁹⁸

These groups of antigens, especially NY-ESO-1 are highly immunogenic, and able to elicit spontaneous humoral and cell mediated immunity when exposed. Naturally this makes an attractive target for immunotherapy in treating cancers, and CTA’s have been found expressed on some solid cancers.¹⁰¹ However, their expression is low or completely absent in AML and MDS as the CTA genes are silenced through promotor hypermethylation.^102,103

Preclinical models in AML/MDS cells treated with HMA had an increase in NY-ESO-1 mRNA through a time and dose-dependent manner through recognition by antigen-specific CD8+ CTL. Circulating myeloid blasts in patients that received HMA also demonstrated de novo NY-ESO-1 expression.¹⁰⁴ Clearly treating MDS patients with epigenetic therapy enhance the expression of proteins once exempt from the immune system. So Griffiths et al. administered an NY-ESO-1 therapeutic peptide vaccine to 7 patients with MDS after 4 cycles of decitabine and observed clinical responses in 5 (3 CR and 2 HI). Since they were all HMA naïve any assertions on efficacy was impossible, however, the authors did observe serially measured immune responses to the vaccine with measurable IFN-γ levels. The immune responses were the most robust in 3 MDS (2CR and 1 SD) patients with detectable baseline NY-ESO-1 specific CD4+ T-cell clones.^104,105

Targeting of antigens exclusive to myelodysplastic progenitor stem cells presents a challenge as many are non-immunogenic within the HSC compartment niche.

The tumor-associated antigens selected for use in therapeutic peptide vaccines described thus far are not unique to MDS. For example, WT-1 is expressed in CLL, CML, AML, primary peritoneal carcinoma, malignant mesothelioma, and many solid cancers to name a few.⁸²

Therefore, WT-1 antigen is not specific to MDS/AML. Rather, it is the disproportionate overexpression of these TAA antigens within their IPSS/IPSSR higher-risk native diseased state relative to their “lower risk” MDS counterparts.

Indeed, across MDS prognostic categories, CD34+ stem cells have WT-1 overexpression with a strict correlation between high mRNA levels and higher blast percentages and IPSS risk.^83,84 Even within the risk subgroups a strong association exists between higher levels of WT-1 mRNA on having increased bone marrow blast percentages and adverse cytogenetics.⁸³

A retrospective study found higher expression of WT1 in the peripheral blood of MDS patients independently predicted a reduced likelihood to obtain a response with HMA, and was associated with an inferior OS that was independent from IPSS-R.⁸⁵

Though the expression of PR3 in HR-MDS has not been fully described in the literature, RNA levels in AML patients were found to be highest with persistent disease rather than adjusted blast counts. Interestingly, PR-3 expression was the highest in favorable ELN risk de novo AML. In contrast, it was the lowest in unfavorable risk ELN cytogenetic risk and in secondary AML from MDS.¹⁰⁶

Whole cell vaccines

Gene transduced tumor cell vaccine: GVAX

Another therapeutic vaccine strategy for MDS used patient derived whole tumor cell (K562) lines transfected with plasmid encoded Granulocyte Macrophage–Colony-Stimulating Factor (GM-CSF) secreting capability.^107–109 The autologous administration of irradiated K562/GM-CSF (GVAX) whole tumor cells primed native dendritic cells to present a panel of malignant myeloid epitopes in an HLA unrestricted manner to activate multiple facets of the adoptive and innate immune system.

A phase I pilot study with GVAX was given as five individual injections to 1 CMML and 4 MDS patients over the course of 3–4 months (Table 3). It was well tolerated with the only grade 1 injection site reactions. There was only 1 major HI without INF gamma production and 1 patient with achieved transfusion independence. Clinical activity was not seen in the three other patients did not have clinical activity though the study was too small to comment on any preliminary efficacy.¹¹⁰

Taken together, therapeutic vaccines for MDS have demonstrated an excellent safety profile but reproducible effectiveness has not yet been achieved despite induction of robust MDS-TAA reactivity in vivo. For a vaccine to be therapeutic it must illicit a strong cytotoxic immune response for stem cell eradication. The initial recognition and processing of vaccine antigen is dependent on innate immune cells and functional antigen presenting cells. A potential issue with relying on the MDS patients’ native APC’s, is the dendritic cell dysfunction inherent to advanced disease. Patients with higher risk MDS have been reported to not only have lower DC counts, but also reduced functionality.¹¹¹

Larger confirmatory and preferably comparator arm trials are necessary to confirm any clinical benefit with vaccination approaches.

In conclusion, several immunotherapeutic approaches that have been effective in other cancers have been attempted in MDS with reported successes in select early phase trials. To date, HSCT remains the only proven effective immunotherapy for MDS, however, the identification and therapeutic modulation of MDS-specific targets such as CD47, TIM-3, WT1, etc. along with combination approaches are likely to be the future of MDS immunotherapy.

HMA-refractory MDS has no available effective treatments and remains the most challenging cohort of MDS patients. Findings from CD47 inhibitor and TIM-3 inhibitor trials are most impressive in MDS patients to date and have outperformed conventional checkpoint inhibitor therapies (PD-1 and CTLA4 inhibition). Single agent and combinatorial approaches with these drugs are now in confirmatory trials and an FDA-approval in this space is closer than ever before.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

Dr. Premal Lulla would like to thank the Edward P. Evans MDS Discovery research grant (PI: Premal Lulla), and The Cancer prevention and research institute of Texas (CPRIT) Early Career Clinical Investigator Award (RP200584).