A novel composite scoring system to predict mortality after auto-HSCT

Currently, there is no composite scoring system to predict mortality following autologous hematopoietic stem cell transplant (auto-HSCT) that takes into account both comorbidities and clinical factors for different diseases.¹ The hematopoietic cell transplant comorbidity index (HCT-CI) comes closest and was used as the backbone for the new Grupo Argentino de Trasplante de Médula Ósea (GATMO) scoring system in a study published by Mariano Berro and colleagues in Biology of Blood and Marrow Transplantation.¹

Methods

The aim of this study was to create a composite scoring system to predict mortality following auto-HSCT and to analyze its association with efficacy outcomes, like overall survival (OS) and non-relapse mortality (NRM). The secondary aim was to analyze the impact of the new scoring system on the following three early morbidity outcomes: the use of mechanical ventilation, dialysis, or vasopressors.

A training cohort of 2,068 adult patients who received auto-HSCT in Argentina was retrospectively identified. The median follow-up of this group was 1.1 years (range, 100 days–14 years) and median transplant year was 2013 (range, 2002−2017). The characteristics of this cohort are shown in Table 1. In brief, 59% of patients were male and just over 50% had multiple myeloma, with 53% being in complete remission. In terms of comorbidities, 58% were considered as low risk with an HCT-CI score of 0.

Table 1. Training cohort characteristics (N = 2,068)¹

HCT-CI, hematopoietic cell transplant comorbidity index
Variable		Patients
Age	Median, years (range)	54 (15−74)
< 55 years, %	52
55−64 years, %	33
≥ 65 years, %	15
Sex, %	Male	59
Disease, %	Multiple myeloma	52
Hodgkin lymphoma	18
Non-Hodgkin lymphoma	30
Pre-transplant chemotherapy lines, %	1	46
2	41
≥ 3	13
Pre-transplant status, %	Complete remission	53
Partial remission	44
Stable/progressive disease	3
HCT-CI score, %	Low risk (0)	58
Intermediate (1−2)	29
High risk (≥ 3)	13

The validation cohort included 2,168 patients who had been treated with auto-HSCT for multiple myeloma or lymphoma either at the Medical College of Wisconsin, US (n = 890) or within the Spanish Group of Hematopoietic Transplantation and Cell Therapy (GETH) (n = 1278). Median follow-up for this group was 1.3 years (range, 100 days–7.5 years).

Results

Training cohort

Early NRM was 3.1%, and long-term NRM at 1 and 3 years was 4.7% and 5.8%, respectively. 1-year OS was 89%, which fell to 65% at 5 years.

Univariate analysis showed that the following variables significantly impacted NRM (the degree of impact was shown by points): male patients (1 point), age (55−64 years = 1 point; ≥ 65 years = 2 points), HCT-CI ≥ 3 (1 point), and disease (Hodgkin lymphoma = 1 point; non-Hodgkin lymphoma = 2 points). These variables were used for the scoring of patients on the GATMO system. Patient risk subgroupings based on the GATMO scoring system are shown in Table 2. The majority of patients fell into the intermediate risk category, with the top two risk categories only accounting for 14% of patients.

The NRM hazard ratio increased proportionally with the GATMO score.

Table 2. Risk stratification groupings and associated GATMO scores¹

Group	Score	Patients (%)
Low risk	0−1	33
Intermediate risk	2−3	53
High risk	4	10
Very high risk	≥ 5	4

The use of mechanical ventilation, vasopressors, or dialysis, as well as early NRM, were significantly associated with the new scoring system (dialysis, p < 0.01; all others, p < 0.001).

The GATMO score was able to significantly differentiate between the four risk groups, in terms of their long-term NRM outcomes at 1 and 3 years and OS outcomes at 1 and 5 years, as shown in Table 3. No significant association was found with relapse rate.

Table 3. GATMO score association with cumulative incidence of NRM and OS in the training cohort¹

CI, confidence interval; GATMO, Grupo Argentino de Trasplante de Médula Ósea; NRM, non-relapse mortality; OR, odds ratio; OS, overall survival; ref, reference. * p values refer to the GATMO score impact on NRM or OS, respectively.
Risk	1-year NRM (%)	3-year NRM (%)	p value*	OR (95% CI)	1-year OS (%)	5-year OS (%)	p value*	OR (95% CI)
Low	1.8	2.3	Ref	—	94	73	Ref	—
Intermediate	3.8	4.9	0.011	2.16 (1.19–3.93)	89	64	0.006	1.43 (1.11–1.84)
High	11.7	14.5	< 0.001	6.43 (3.33–12.41)	76	48	< 0.001	2.54 (1.79–3.60)
Very high	25.0	27.4	< 0.001	12.80 (6.29–26.04)	65	52	< 0.001	3.99 (2.60–6.13)

Validation cohort

The validation cohort was used to confirm the results from the training cohort. This group had a median age of 60 years (range, 15−81) and 60% were male; 61% of patients had multiple myeloma, 31% had non-Hodgkin lymphoma, and 8% had Hodgkin lymphoma. In terms of comorbidities, 16% were considered as low risk with an HCT-CI score of 0. Similar to the training cohort, the GATMO score was significantly associated with the three early morbidity outcomes (vasopressor use, dialysis, and mechanical ventilation), as well as NRM at 1 and 3 years and OS at 1 and 5 years, as shown in Table 4. These results further confirmed the significant association of both endpoints with the GATMO score.

Table 4. GATMO score association with long-term NRM and OS in the validation cohort¹

CI, confidence interval; GATMO, Grupo Argentino de Trasplante de Médula Ósea; NRM, non-relapse mortality; OR, odds ratio; OS, overall survival; ref, reference. * p values refer to the GATMO score impact on NRM or OS, respectively
Risk	1-year NRM (%)	3-year NRM (%)	p value*	OR (95% CI)	1-year OS (%)	5-year OS (%)	p value*	OR (95% CI)
Low	0.9	3.1	Ref	—	96	81	Ref	—
Intermediate	2.2	5.8	0.030	2.38 (1.08–5.23)	93	68	0.018	1.56 (1.08–2.25)
High	4.7	8.2	0.002	3.78 (1.64–8.69)	88	57	< 0.001	2.98 (1.060–3.59)
Very high	8.5	11.2	< 0.001	5.74 (2.39–13.77)	81	60	< 0.001	3.04 (0.93–4.79)

Conclusion

The results of this retrospective study indicate that variables other than comorbidities affect mortality post auto-HSCT. More specifically, by combining HCT-CI score along with age, sex, and disease in a new, composite GATMO score, the authors we able to predict early morbidity events, as well as NRM and OS. This composite score allows for a new stratification of patients according to risk, which may help with patient transplant eligibility and the adjustment of treatment intensity.