Allogeneic Hematopoietic Stem Cell Transplantation is Underutilized in Older Patients with Myelodysplastic Syndromes

Abstract

Allogeneic hematopoietic stem cell transplantation (HCT) is the only curative treatment for myelodysplastic syndrome (MDS). The proportion of MDS patients referred for transplant evaluation, those undergoing transplantation and the reasons for not undergoing transplant are unknown. In this retrospective analysis, pre-defined HCT eligibility and indications criteria were applied to 362 unselected patients with newly diagnosed MDS seen by Leukemia faculty between 2008 and 2015 at Memorial Sloan Kettering Cancer Center. Two hundred ninety four patients (81%) were deemed eligible for transplant and among these, transplant was considered indicated in 244 (83%). Of these, 158/244 (65%) were referred for transplant evaluation at a median of 3.9 months from diagnosis. Overall 120/362 (33%) underwent transplant at a median of 7.7 months from diagnosis. Metastatic solid organ malignancy was the major reason for transplant ineligibility (54%), and death due to MDS, which occurred in 41% of candidates who were not transplanted, was the major reason for not undergoing transplant. Factors associated with a lower likelihood of referral for transplant evaluation included age ≥65 (p<0.001), >2 co-morbidities (p=0.008), intermediate-1/low risk MDS (p<0.001), <5% blasts at diagnosis (overall p<0.001), having medicare/medicaid health insurance (p<0.001), not being married (p=0.017) and diagnosis between 2008-2011 (p=0.035). On multivariate analysis adjusting for all of the previous, diagnosis between 2008-2011 (p<0.001), age ≥65 (p=0.001) and <5% blasts at diagnosis (overall p=0.031) were associated with a lower likelihood of referral for transplant evaluation. Factors associated with a lower likelihood of undergoing transplant included age ≥65 (p<0.001), ≥2 co-morbidities (p=0.003), intermediate-1/low risk MDS (p<0.001), <5% blasts (overall p<0.001), very low/low/intermediate risk IPSS-R karyotype (p=0.018) and having medicare/medicaid health insurance (p<0.001). In multivariate analysis adjusting for all of the previous, age ≥65 (p=0.021), presence of ≥2 comorbidities (p=0.018) and <5% blasts (overall p=0.011) were associated with a lower likelihood of undergoing transplant. The results highlight that transplantation for MDS remains underutilized particularly for candidates over the age of 65.

Introduction

In the era of hypomethylating agents, data regarding the proportions of MDS patients referred for transplant evaluation and proceeding to transplant as well as the barriers to transplant are unknown. In 2007, Estey et al reported on a cohort of 99 patients with high-risk MDS and AML of whom only 53 underwent transplant evaluation and 14 underwent reduced-intensity transplant¹. In 41% of patients the reason for not receiving a transplant was “unclear”.

Mathematical models which showed that higher risk MDS patients benefited from early transplant are likely to have influenced patterns of transplant referral^2,3; however, factors associated with referral have never been assessed. Transplantation practices have evolved since the publication of these models with increased use of haploidentical and HLA mismatched donors, reduced intensity conditioning regimens², novel GVHD prophylaxis strategies and improved supportive care^3,4. While highly informative, these studies may have resulted in bias to delay referral of lower risk (by IPSS) candidates, potentially impacting their likelihood to undergo transplant. The contemporary definition of high-risk MDS now includes improved cytogenetic stratification provided by the IPSS-R and data showing that somatic mutation profiling can predict outcomes of non-transplant⁵ and transplant therapy^6,7. In this context, reliance on IPSS as a sole surrogate to define high-risk patients and guide the decision on when to undertake transplant may have inadvertently biased the timing of referral for transplant evaluation.

Despite making up the majority of all newly diagnosed MDS patients, data suggest that patients ≥65 account for a minority (7.5%) of transplanted MDS patients^8,9. Comorbidities and disease risk rather than chronologic age are consistently associated with transplant outcomes¹⁰ and on this basis, medical reimbursement criteria for transplant for patients over 65 were updated by the Center for Medicare and Medicaid Services (CMS) to allow older patients access to transplant as part of a clinical trial¹¹.

Socioeconomic variables including income, racial minority, level of education and insurance status are associated with lower transplant rates for leukemia and lymphoma, but have never been reported in MDS¹². The socioeconomic factors that impact access to HCT for older MDS patients are unknown, but likely center around patterns of referral for HCT evaluation, which vary between private and academic practice.

We conducted a retrospective analysis of newly diagnosed MDS patients seen at MSKCC to determine the fraction of patients referred for transplant, and to study the factors associated with patients proceeding to transplant. We defined transplant eligibility based on organ function and performance status and considered ‘patient’ variables including Karnofsky Performance Status (KPS), number of comorbidities and socioeconomic factors together with MDS ‘disease’ variables including blasts count, IPSS and karyotype according to IPSS-R to define groups at risk of non-referral for transplant evaluation and those having a lower likelihood of undergoing this potentially curative procedure.

Materials and Methods

Patients and definitions

This was a single center retrospective study. To minimize selection bias, only patients seen first by a Leukemia Service faculty physician at our institution within 6 months of the MDS diagnosis between January 2008 and December 2015 were included. To limit analysis to transplant candidates, only those aged ≤75 years at diagnosis were included. Diagnostic bone marrow histology was available in all patients, complete blood count parameters in 99.7% (n=361) and cytogenetics in 98.6% (n=357). IPSS and IPSS-R were calculated for 96.7% patients (n=350). Fourteen co-morbidities were specifically defined (supplementary table 1) and catalogued from clinical records at the time of MDS diagnosis. Therapy-related MDS was defined as MDS arising in patients with any prior chemotherapy or radiotherapy exposure. Patient reported socioeconomic variables were sourced from the clinical database and included insurance and employment status, ethnicity, religion and resident zip code. HCT graft sources included matched related and unrelated donors, mismatched unrelated, cord blood and haploidentical grafts.

High resolution HLA typing of patient and sibling samples was performed through the American Red Cross (turn-around time of 5 business days). A preliminary unrelated donor search (turn-around time 2 working days) was performed using the Haplogic system¹³ This preliminary unrelated search is rapid and highly accurate at predicting the likelihood of identifying an unrelated donor and includes an assessment of suitable cord blood units.

Post transplant care is provided for all patients at our center for the first 3 months after HCT and patients are expected to live within 1 hour of the transplant facility with a caregiver. Thereafter patients whose primary place of residence is out of area are likely to return home if medically stable. They continue to be followed primarily at MSK; however, interim visits may be with a local provider. Accommodation is provided free of charge for ‘out of area’ patients for the first 3 months after HCT to facilitate optimal post HCT care.

Definition of transplant eligibility and indication

Patients were considered ineligible for HCT if at the time of first assessment by the Leukemia service they had a history of metastatic solid organ malignancy within 5 years of MDS diagnosis, documented left ventricular ejection fraction <40%, creatinine >2.0 mg/dL, documented FEV1 <1.5L or DLCO adjusted for hemoglobin <50%, KPS <60 (ECOG ≥3), documented liver cirrhosis or active uncontrolled infection. Prior hematologic malignancy did not impact eligibility nor did age as all candidates were ≤75. Cardiorespiratory parameters (ejection fraction, FEV1 and DLCO) were included if results were available at the time of first leukemia visit. These criteria were assessed retrospectively in all patients at the time of the first referral to the Leukemia faculty and were agreed upon by both leukemia and transplant faculty as contraindications for transplantation.

Factors used to determine if transplant was indicated were kept constant across the study period and included validated high-risk disease features. At the time of MDS diagnosis transplant was indicated if patients had excess blasts (≥5%)¹⁴, IPSS score Intermediate-2 or high^15,16, therapy related MDS^16,17, life threatening cytopenia^15,18 (hemoglobin <7g/dL, absolute neutrophils <0.5×10^9/L or platelets <30×10^9/L) or high/very high risk cytogenetics¹⁹ (defined by IPSS-R cytogenetic risk). Patients who did not meet these criteria at diagnosis but who had progressive disease defined by blasts ≥5% were considered eligible from the time of disease progression. Similar criteria have been applied by others²⁰.

Reasons for non-transplant

In eligible patients who had an indication for HCT but did not undergo transplant a single reason for non-transplant was attributed in the following order of priority defined at the time of last follow-up: death, uncontrolled disease progression, new co-morbidity precluding transplant, patient refusal, documented lack of necessary social support, lack of suitable donor, no-transplant referral and loss to follow-up.

Statistical analysis

Analyses relating to the time to transplant referral and time to transplant were done within the competing risks framework. In order to focus the analysis of time to transplant referral and time to transplant to patients who were candidates for transplant, all 294 transplant eligible patients at diagnosis were included in both analyses, with date of diagnosis considered to be time zero. Deaths prior to transplant referral, or transplant were considered competing risks for the time to transplant referral and time to transplant analyses. Univariate and multivariate cause-specific Cox proportional-hazards regression models were used to evaluate the effect of patient and disease related variables on the time to transplant referral, and time to transplant. Differences in diagnosis and pre-transplant laboratory parameters were determined using the Wilcoxon rank-sum test for paired data.

Results

Cohort

362 patients with a median (range) age at diagnosis of 65 years (20-75) and median follow-up of 17 (0-94) months in those alive at last review were included (table 1). Men (64%) made up a larger proportion of patients. Therapy-related MDS was diagnosed in 36% of patients and 203 (56%) had excess (≥5%) blasts at the time of diagnosis. Of the 158 (44%) patients with <5% blasts at diagnosis 34 (22%) progressed to excess blasts while 77/362 (21%) of all patients progressed to acute myeloid leukemia (AML).

Table 1:

Characteristics of all 362 patients included in the analysis

	N (%)
Total	362
Patient variables
Year of MDS diagnosis
2008-2009	51 (14.1)
2010-2011	116 (32.0)
2012-2013	100 (27.6)
2014-2015	95 (26.2)
Median Age at diagnosis (range)	65.6 (20.4-75.8)
Age groups
<40	22 (6.1)
40-64	169 (46.7)
≥65	171 (47.2)
Gender
Men	233 (64.4)
Women	129 (35.6)
Number of comorbidities at diagnosis
0	82 (22.7)
1	114 (31.5)
2	81 (22.4)
≥3	85 (23.5)
KPS at diagnosis
<90%	75 (20.7)
≥90%	263 (72.7)
Missing data	24 (6.6)
Disease variables
MDS type
De Novo MDS	231 (63.8)
Therapy related MDS	131 (36.2)
Prior hematologic cancer	51 (38.9)
Prior solid organ cancer	77 (58.8)
Other cytotoxic exposure	3 (2.3)
Hypocellular MDS
Hypocellular MDS	20 (5.5)
Non hypocellular MDS	342 (94.5)
Fibrosis
MDS with Grade 2 or 3 fibrosis	50 (13.8)
Non Fibrotic MDS	312 (86.2)
Blast at diagnosis
<5%	158 (43.6)
5-9%	92 (25.4)
≥10%	111 (30.7)
Missing data	1 (0.3)
IPSS calculated
Low	46 (12.7)
Int-1	119 (32.9)
Int-2	140 (38.7)
High	45 (12.4)
Missing data	12 (3.3)
IPSS-R
Very Low	26 (7.2)
Low	72 (19.9)
Intermediate	83 (22.9)
High	77 (21.3)
Very High	99 (27.3)
Missing data	5(1.4)
Socioeconomic variables
Ethnicity
White	292 (80.7)
Hispanic	15 (4.1)
Black	23 (6.4)
Asian	16 (4.4)
Missing data	16 (4.4)
Religion
Christian	229 (63.3)
Jewish	62 (17.1)
Other/None identified	67 (18.5)
Missing data	4 (1.1)
Married
Yes	260 (71.8)
No	102 (28.2)
State
New York	215 (59.4)
New Jersey	59 (16.3)
Other	88 (24.3)
Distance (km) to MSKCC from home*
0.5 – 20.8	91 (25.1)
20.9 – 46.1	91 (25.1)
46.2 – 127.2	90 (24.9)
127.3 – 12530.0	90 (24.9)
Insurance**
Medicare/Medicaid	157 (43.3)
Private/Commercial/Self-pay	205 (56.6)
Household Income above US median#
Yes	237 (66.9)
No	117(33.1)
Employed at MDS diagnosis
Yes	324 (89.5)
No	38 (10.5)
Type of work^
White collar	193 (53.3)
Blue Collar	47 (13.0)
NA/Missing data	122 (33.7)
Donors used for patients undergoing HCT
Matched unrelated	47 (39.2)
Matched related	42 (35.0)
Mismatched unrelated	19 (15.8)
Cord Blood	9 (7.5)
Mismatched related¥	3 (2.5)

^*Median distance from home to MSKCC for entire population was 46.12 Km, hence this value was used to separate the population.

^#Data sourced from US Census Bureau statistics (2012) accessed at http://zipwho.com/ (accessed September 10th 2016 at 8pm). The median annual household income by ZIP for included patients was 60,245$ (16,664-153,632). This excludes international patients.

^{^}Blue and white-collar employment were defined per US government published definitions. See: https://www.opm.gov/policy-data-oversight/data-analysis-documentation/federal-employment-reports/reports-publications/the-twenty-largest-white-collar-occupations/ and https://www.opm.gov/policy-data-oversight/data-analysis-documentation/federal-employment-reports/reports-publications/the-twenty-largest-blue-collar-occupations/

^**Insurance type at the time of First evaluation at MSKCC

^¥Of the mismatched related donors 2 were classed as haploidentical.

Eligibility and indications for transplant

Of the entire cohort, 294 (81%) patients met transplant eligibility criteria and 120 (33%) were transplanted (figure 1). The most common reasons for transplant ineligibility were a history of metastatic solid organ malignancy within 5 years of MDS diagnosis (54%), inadequate organ function (32%) and KPS<60 at presentation (13%) (table 2). Of those eligible for transplant, 244 (83%) patients had at least one of the previously defined transplant indications. Only 158/244 (65%) of patients with at least one indication for transplant were referred for transplant evaluation and 109/158 (69%) of referred patients underwent transplant (Figure 1). The primary reasons for eligible patients who had a transplant indication (n=244) to not undergo transplant (n=135) included death during follow-up (44%) and disease progression (17%). No reason could be attributed to 24% that were lost to follow-up. Ten (16%) developed a new comorbidity that prohibited them from undergoing transplant (table 2). None of the ineligible patients underwent transplant; however, 11 patients without a transplant indication as defined in our study underwent transplant at the discretion of the transplant and leukemia service physicians.

Figure 1.

Clinical progression of all patients.

All 362 patients were initially classified by their eligibility to undergo transplant and subsequently by whether they had an indication to undergo transplant according to predefined criteria. The number of patients referred for transplant and the number who underwent transplant in each group was assessed.

Table 2:

Reasons for transplant ineligibility and non-transplant

Reasons for BMT Ineligibility at MDS diagnosis	N (%)
Total	68 (18.8)
Metastatic Cancer within 5 years of MDS diagnosis	37 (54.4)
ECOG >2	9 (13.2)
DLCO adjusted for Hemoglobin <50% or FEV1<1.5L	7 (10.3)
Left ventricular ejection fraction <40%	6 (8.8)
Creatinine >/=2mg/dL	5 (7.4)
Documented liver cirrhosis	4 (5.9)
Uncontrolled infection	0
Reason for non transplant among 135 patients who were eligible and who had a transplant indication.
Total	135
Died*	60 (44.4)
Lost to follow-up	32 (23.7)
Progressive MDS at last review	20 (14.8)
New co-morbidity, still alive**	10 (7.4)
Refused	6 (4.4)
No suitable donor	4 (3.0)
Not referred for HCT	2 (1.5)
Lack social support	1 (0.7)

DLCO: Oxygen diffusion capacity, FEV1: forced expiratory volume in 1 second.

^*Cause of death was attributed to MDS in 56/60 (93%) patients who died and was unknown in 4/60 (7%).

^**Development of a new co-morbidity in those alive at last follow-up was the primary reason for non-transplant in 10 (7.4%) patients. These new co-morbidities included organ failure (n=3), other active neoplasm (n=3), uncontrolled infection (n=2), patient reaching age >75yr (n=1) and development of other co-morbidity (n=1).

Factors associated with non referral for transplant evaluation

Among transplant eligible patients (n=294), factors associated with non referral for HCT evaluation grouped into patient or disease variables were assessed in univariate (table 3) and multivariate analysis (table 4). In univariate analysis ‘patient’ variables associated with a lower likelihood of referral included: MDS diagnosis between 2008-2011 (vs 2012-2015, p=0.035), age at diagnosis ≥65 (p<0.001), presence of ≥2 co-morbidities at diagnosis (p=0.008), having medicare/medicaid health insurance (p<0.001) and not being married (p=0.017). Low-risk disease features were also associated with a lower likelihood of referral including intermediate-1/low-risk IPSS (p<0.001) and absence of excess blasts (overall p<0.001). In multivariate analysis (table 4), adjusted for all of the previously significant univariate variables, year of MDS diagnosis between 2008-2011 (p<0.001), age ≥65 (p=0.001) and no excess in blasts (overall p=0.031) remained independently associated with a lower likelihood of being referred for transplant evaluation.

Table 3:

Univariate analysis of factors associated with referral for transplant evaluation and transplant status^{^}

		Time to transplant referral			Time to transplant
	N	Referred	HR (95% CI)	P-value	Transplanted	HR (95% CI)	P-value
Total evaluated	294	179			120
Year of MDS diagnosis				0.035			0.150
2012-2015	156	91	1		57	1
2008-2011	138	88	0.726 (0.539, 0.978)		63	0.765 (0.531, 1.102)
Age at diagnosis				<0.001			<0.001
<65	152	114	1		85	1
≥65	142	65	0.490 (0.361, 0.664)		35	0.428 (0.288, 0.634)
Gender				0.984			0.472
Female	100	64	1		48	1
Male	194	115	0.997 (0.734, 1.355)		72	0.874 (0.606, 1.261)
Total comorbidities				0.008			0.003
<2	192	128	1		93	1
≥2	102	51	0.645 (0.466, 0.894)		27	0.525 (0.342, 0.806)
Diagnosis KPS				0.604			0.134
<90	43	24	1		10	1
≥90	232	146	0.892 (0.578, 1.375)		105	1.645 (0.859, 3.150)
Coronary artery disease				0.001			0.005
No	250	164	1		113	1
Yes	44	15	0.415 (0.244, 0.705)		7	0.332 (0.155, 0.714)
Diabetes				<0.001			0.001
No	250	164	1		114	1
Yes	44	15	0.367 (0.216, 0.624)		6	0.242 (0.106, 0.549)
Cerebrovascular disease				0.021			0.048
No	278	174	1		118	1
Yes	16	5	0.350 (0.144, 0.852)		2	0.244 (0.060, 0.986)
IPSS				<0.001			<0.001
Int-2/high	143	94	1		69	1
Low/Int-1	141	82	0.577 (0.427, 0.779)		49	0.456 (0.313, 0.663)
Diagnosis blast %				<0.001			<0.001
≥10	92	60	1		44	1
5-9	77	53	0.940 (0.649, 1.360)		37	1.057 (0.682, 1.637)
<5	125	66	0.502 (0.352, 0.714)		39	0.442 (0.286, 0.683)
IPSS-R karyotype				0.172			0.018
High/very high	95	55	1		40	1
Very low/low/Int	196	124	0.800 (0.580, 1.102)		80	0.627 (0.425, 0.923)
Therapy related MDS				0.540			0.706
No	210	128	1		86	1
Yes	84	51	1.107 (0.800, 1.532)		34	1.080 (0.725, 1.607)
Insurance				<0.001			<0.001
Private	166	124	1		90	1
Medicare	117	49	0.462 (0.332, 0.644)		28	0.429 (0.280, 0.655)
Distance to MSK				0.246			0.704
<Median	140	87	1		61	1
>Median	154	92	1.190 (0.887, 1.596)		59	1.072 (0.749, 1.533)
Ethnicity				0.397			0.631
Non-white	44	28	1		16	1
White	235	145	0.839 (0.559, 1.259)		101	1.139 (0.671, 1.933)
Income				0.925			0.604
<Median	96	60	1		39	1
>Median	190	113	1.015 (0.742, 1.389)		79	1.107 (0.754, 1.625)
Religion				0.364			0.665
Christian	187	119	1		79	1
Jewish	52	31	0.808 (0.543, 1.203)		21	0.852 (0.525, 1.382)
Other/none identified	53	29	0.791 (0.527, 1.188)		20	0.830 (0.508, 1.358)
Married				0.017			0.098
Yes	217	141	1		94	1
No	77	38	0.645 (0.450, 0.924)		26	0.692 (0.448, 1.070)
State				0.699			0.940
NJ	42	20	1		13	1
NY	173	112	1.138 (0.707, 1.834)		78	1.100 (0.611, 1.981)
Other	79	47	1.247 (0.738, 2.106)		29	1.119 (0.582, 2.153)
Employed				0.703			0.786
No	32	18	1		15	1
Yes	262	161	1.100 (0.675, 1.793)		105	0.928 (0.539, 1.596)
Occupation #				0.292			0.204
Blue collar	38	25	1		16	1
White collar	157	116	1.262 (0.819, 1.944)		86	1.415 (0.829, 2.416)

^{^}Cause specific hazard model, death was regarded as a competing risk. Only patients who were eligible for transplant were included in the analysis. Date of MDS diagnosis was regarded as time zero.

^#Occupation type was identified in 195 patients

NJ: New Jersey, NY: New York

Table 4:

Multivariate analysis of factors associated with referral for transplant evaluation and transplant.

Variable	N	Number referred for transplant evaluation	HR (95% CI)	P-value
Year of MDS diagnosis				<0.001
2012-2015	139	84	1
2008-2011	135	86	0.515 (0.369, 0.717)
Patient age (years)				0.001
<65	144	110	1
≥65	130	60	0.459 (0.295 0.715)
Diagnosis blast %				0.031
≥10	87	58	1
5-9	75	52	0.955 (0.626, 1.456)
<5	112	60	0.571 (0.346, 0.942)
IPSS				0.106
Intermediate 2/high	139	92	1
Low/intermediate 1	135	78	0.705 (0.461, 1.078)
Total comorbidities				0.212
<2	180	122	1
≥2	94	48	0.801 (0.565, 1.135)
Insurance				0.393
Private	162	122	1
Medicare	112	48	0.818 (0.517, 1.296)
Married				0.063
Yes	204	134	1
No	70	36	0.702 (0.483, 1.019)
Variable	N	Number transplanted	HR (95% CI)	P-value
Patient age (years)				0.021
<65	144	83	1
≥65	130	33	0.543 (0.323, 0.913)
Diagnosis Blast %				0.011
≥10	87	42	1
5-9	75	37	1.404 (0.847, 2.327)
<5	112	37	0.654 (0.360, 1.188)
IPSS				0.078
Intermediate 2/high	139	68	1
Low/intermediate 1	135	48	0.610 (0.351, 1.058)
IPSS-R karyotype				0.185
High/very high	90	40	1
Very low/low/intermediate	184	76	0.734 (0.465, 1.159)
Total comorbidities				0.018
<2	180	90	1
≥2	94	26	0.579 (0.368, 0.911)
Insurance				0.372
Private	162	88	1
Medicare	112	28	0.776 (0.444, 1.355)

Specific comorbidities associated with a lower likelihood of referral included coronary artery disease (p=0.006), diabetes (p=0.001) and cerebrovascular disease (p=0.020), all of which were significantly associated with the time to transplant referral, after adjusting for patient age (</≥65) at diagnosis.

Factors associate with not undergoing transplantation

Factors associated with not undergoing transplant were evaluated among eligible candidates (n=294) (table 3). On univariate analysis, age at diagnosis >65 (p<0.001), having ≥2 comorbidities (p=0.003), and not having medicare/medicaid health insurance (p<0.001) were associated with a lower likelihood of undergoing transplant. Disease risk variables associated with a lower likelihood of transplantation included: intermediate-1/low risk IPSS (p<0.001), no excess of blasts at diagnosis (overall p<0.001) and very low/low/intermediate risk cytogenetics (defined according the IPSS-R classification) at the time of diagnosis (p=0.018). In multivariate analysis (table 4) adjusted for all of the previously significant univariate factors, age ≥65 (p=0.021), presence of ≥2 comorbidities (p=0.018) and absence of excess blasts (overall p=0.011) remained independently associated with a lower likelihood of undergoing transplant.

Specific comorbidities associated with a lower likelihood of undergoing transplant included coronary artery disease (p=0.017), diabetes (p=0.003) or cerebrovascular disease (p=0.039), all of which remained significant after adjusting for age (</≥65) at diagnosis.

Pre transplant therapy

Out of the 120 patients who underwent HCT, 112 (93%) received diseasemodifying therapies including: hypomethylating agent (HMA) only (n=70, 58%); HMA and induction chemotherapy (n=28, 23%); induction chemotherapy alone (n=11, 9%) and other therapies in (n=3, 3%). The remaining 8 patients (7%) received only supportive care. Patients who underwent transplant were noted to have significant changes in physiologic and biochemical parameters between the time of diagnosis and transplant. Bilirubin decreased by a median of 0.1mg/dL (p<0.001), serum albumin decreased by a median of 0.2g/dL (p<0.001) and KPS decreased by at least 10 points for 42% of patients (p<0.001). Weight (p=0.069) and glomerular filtration rate (p=0.089) remained unchanged (Supplementary table 2).

Discussion

In a cohort of 362 consecutive patients referred to a large academic center for a diagnosis of MDS in the era of HMA therapy, the majority (81%) of patients were considered eligible for transplant. Of these, 61% were referred for transplant evaluation, while only one third of all patients were transplanted. Patients older than 65 years were less likely to be referred for transplant evaluation and were less likely to undergo transplant independently of disease risk, performance status and comorbidity (figure 2). Despite measures to improve transplant access for older patients, age remains a significant barrier to what is the only curative therapy for this disease.

Figure 2. Time to transplant referral and time to transplant by age group.

Cumulative incidence of time to transplant referral (A) and time to transplant (B) stratified by patient age. Among transplant eligible patients (n=294) patients aged ≥65 were less likely to be referred for transplant (p<0.001) and were less likely to undergo transplant (p<0.001).

In a large cross sectional survey study conducted in the US from 2005 to 2007, Sekeres et al reported that only 4% of newly diagnosed MDS patients had been referred or were considered for transplant referral; however, only 10-20% of surveyed hematologists had a practice based at a comprehensive cancer center or university hospital²¹. The percentage of patients referred for transplant was vastly different from the rate of transplant referral in the present study. The difference may be accounted for by varied referral patterns in private and academic practice and amongst different socioeconomic and geographic regions^22,23. The present study was conducted out of an academic cancer center where a streamlined referral process between the leukemia and transplant services facilitated early referral. Based on socioeconomic factors, patients included in this cohort were not representative of the US median as most were privately insured (57%), employed (90%), identified themselves as white (81%) and resided in areas with household incomes above the US median (67%) (table 1). All these variables have been associated with greater access to transplant in prior studies of non-MDS patients²⁴. Sekeres et al did not evaluate patient socioeconomic variables so these could not be compared with patients in this study. Thus, reported underutilization of HCT and referral for transplant evaluation appears to be significantly more pronounced in private practice and possibly among less affluent populations of patients with MDS.

The ideal timing of transplantation for MDS in the HMA era is unknown. Early publications suggest a benefit for transplant in patients with intermediate-2/high-risk IPSS ^2,3 while more recent reports indicated a benefit for patients with intermediate-1/2 or high IPSS-risk disease²⁵. Importantly, the intention of these studies was not to guide timing of referral for transplant evaluation but rather to assist transplant physicians in deciding the appropriate time to transplant patients primarily based on disease risks. Factors influencing referral for transplant evaluation are unknown. We hypothesized that higher risk patients were referred earlier, as non-transplant clinicians were likely influenced by reports identifying a survival benefit with transplant only in higher risk patients. Indeed, in this cohort, patients with intermediate-2 or high-risk IPSS and those with excess blasts were more likely to be referred. The analysis demonstrated that among candidates, the most common reason for non-transplant was disease related mortality. On this basis and given that the ideal timing of transplant in the current era remains unknown we argue that all eligible candidates should be referred for transplant evaluation close to the time of diagnosis irrespective of disease risk. This would facilitate early discussion regarding transplant and timely procurement of donors.

Administration of chemotherapy to patients with advanced MDS prior to transplant continues to be debated ^26–29. In this analysis, we found significant physiologic changes in patients between diagnosis and the time of transplant; 16% of eligible patients acquired comorbidities, which made them ineligible for transplant. Moreover, among those transplanted, significant changes in hepatic function, albumin and performance status were identified between the time of MDS diagnosis and the time of transplant. These changes occurred during the administration of disease modifying therapy and supportive care, raising the concern that prolonged pre-transplant periods adversely impact transplant outcomes as patients acquire co-morbidities, some of which may influence transplant candidature and outcome³⁰. Although hepatic function, hypoalbuminemia and low performance scores are associated with adverse transplant outcomes³¹, the clinical significance of changes identified here needs further study. None-the-less, we feel the data strengthen the argument for early referral of all candidates.

Patients diagnosed with MDS before 2012 were less likely to be referred for transplant. This may represent an inflection point in 2011 when Medicare began to cover allogeneic transplant for those >65 under the ‘Coverage with Evidence Development’ (CED) mechanism. However, despite this change, patients aged >65 remained less likely to undergo transplant independent of co-morbidity and disease risk. This may be a attributed to limited prospective data on transplant outcomes among elderly MDS patients and data showing improved survival in those with IPSS intermediate-2/high risk MDS treated with azacitidine³². This may have resulted in hesitancy to refer patients for transplant evaluation with a preference to administer hypomethylating agents. The benefit of transplant in cohorts comprising older patients is being prospectively studied in a multi-center biologic assignment trial comparing reduced intensity HCT to hypomethylating therapy or best supportive care in patients aged 50-75 with intermediate-2 and high-risk MDS³³. While this is a progressive step, the study is only including patients with available 8/8 HLA matched donors thus leaving unanswered questions regarding the benefit of alternate donor transplant versus non-transplant strategies in older MDS patients.

We found that individuals with vascular disease and diabetes were less likely to be referred for transplant evaluation. The HCT-CI score is validated to predict transplant-related mortality in patients with MDS^34,35. Specific comorbidities are credited with weighted scores, with 3 points attributed to severe pulmonary and hepatic dysfunction, cardiac valve disease and prior malignancy while diabetes, cerebrovascular and coronary disease are attributed a single point³⁴.We catalogued 14 patient comorbidities at the time of MDS diagnosis using clinical criteria (supplementary table 1). Among eligible patients, those with >2 comorbidities at diagnosis as well as those with vascular disease (diabetes, cerebrovascular or coronary artery disease) were less likely to undergo transplant independently of their age. This highlights that comorbidities associated with reduced likelihood of undergoing transplant and transplant referral are not necessarily the same as validated factors that predict transplant outcomes. Therefore, the comorbidity score calculated at the time of MDS diagnosis could have potential utility in promoting earlier transplant referral.

This study has several limitations, mainly its retrospective design and single specialty cancer institution setting. An important point to highlight in this context is graft source; all transplant donor sources were offered including cord blood and haploidentical transplantation and as a result absence of a suitable donor was not identified as a major barrier to HCT. At centers that have less experience in alternate donor transplantation this is likely to be a greater limitation, and therefore we believe that donor/graft source maybe a major barrier in the community. This cohort had a large proportion of patients with therapy related MDS as many patients were referred internally. This impacts reasons for HCT ineligibility, which are likely to vary in proportions if assessed at other centers. Finally, the socioeconomic status of patients included in the study may not reflect of the national or international experience as well the experience of practitioners in private practice.

These data highlight the disparity between the number of patients who are eligible for transplant and the fraction that undergoes this therapy. Patients aged >65 were less likely to undergo transplant due to a lower rate of referral for transplant evaluation, death from disease progression and development of a comorbidities. Given the lack of potentially curative non-transplant MDS therapies, all newly diagnosed patients aged ≥75 should be referred for transplant evaluation at diagnosis, irrespective of disease risk. Early referral may increase the likelihood of transplant by minimizing complications acquired before HCT. Transplant should be considered for every patient with MDS, but the decision to proceed requires careful consideration of the potential benefits and risks.

Highlights:

• Approximately 80% of MDS patients below the age of 75 are eligible to undergo transplant.

• Only one third of all MDS patients referred to an academic center undergo transplant

• Death due to MDS and disease progression are the major barriers to transplant.

• Patients aged over 65 were less likely to be referred for transplant evaluation and were less likely to undergo transplant.