Overall survival of glasdegib in combination with low-dose cytarabine, azacitidine, and decitabine among adult patients with previously untreated AML: comparative effectiveness using simulated treatment comparisons

Figures & data

Table 1 Baseline characteristics of selected studies

Reference	Intervention	N	Median Age	Male, n (%)	De novo, n (%)/Secondary, n (%)	Median hemoglobin, g/dL (range)	ECOG PS 0/1, n (%)	Bone Marrow Blasts >50% n (%)	Cytogenetic risk (poor), n (%)
Cortes 2016 (Individual Patient Data)^8	GLAS + LDAC: GLAS 100 mg orally, once daily + LDAC 20 mg SC twice daily, days 1–10 of each 28-day cycle	78	77	69 (75.6)	38 (48.7)/40 (51.3)	9.1 (6.4–14.0)	37 (47.4)	31 (39.8)	25 (32.1)
	LDAC: 20 mg SC twice daily, days 1–10 of each 28-day cycle	38	76	23 (60.5)	18 (47.4)/20 (52.6)	9.3 (6.0–14.6)	20 (52.6)	18 (47.4)	16 (42.1)
Dombret 2015^5	AZA: 75 mg/m^2 SC per day for 7 consecutive days per 28-day cycle for ≥6 cycles	241	75	139 (57.7)	192 (79.7)/49 (20.3)	9.5 (5.0–13.4)	186 (77.2)	173 (71.8)	85 (35.3)
	LDAC: 20 mg twice daily for 10 days per 28-day cycle for ≥4 cycles	158	75	94 (59.5)	135 (85.4)/23 (14.6)	9.3 (5.6–14.4)	123 (77.9)	128 (81.0)	54 (34.2)
Kantarjian 2012^6	DEC: 20 mg/m^2 once daily for 5 consecutive days every 4 weeks	242	73	137 (56.6)	155 (64.0)/87 (36)	9.3 (5.2–15.0)	184 (76.0)	105 (43.4)	87 (36.1)
	LDAC: LDAC 20 mg/m^2 SC once daily for 10 consecutive days every 4 weeks	215	73	131 (60.9)	140 (65.1)/73 (34.0)	9.4 (5.0–12.6)	164 (76.3)	90 (41.9)	79 (36.9)

Abbreviations: AZA, azacitidine; DEC, decitabine; ECOG PS, Eastern Cooperative Oncology Group Performance Status; GLAS, glasdegib; LDAC, low-dose cytarabine; SC, subcutaneously.

Figure 1 Comparison network.

Notes: In the above comparison network, LDAC alone is the common comparator between trials. In the GLAS + LDAC versus LDAC (Cortes 2016) trial and AZA versus LDAC (Dombret 2015) trial, LDAC was administered as 20 mg twice per day . In the DEC vs LDAC (Kantarjian 2012) trial, LDAC was administered as 20 mg/m² once daily . Either dose schedule is considered to have comparable drug concentration over time (area under the curve) which includes any associated cytotoxic effects.¹³
Abbreviation: AML, acute myeloid leukemia.

Figure 2 Multi-stepped criteria to conduct and evaluate simulated treatment comparisons.

Abbreviations: AFT, accelerated failure time; AIC, Akaike’s information criterion; BIC, Bayesian information criterion.

Table 2 Variable selection: GLAS + LDAC vs AZA

Full Model Results			Statistical Evidence	Justification for Inclusion in Stepwise Models
Included Baseline Characteristics	GLAS + LDAC (IPD)	AZA (Dombret 2015)	GLAS + LDAC versus LDAC IPD Cox p-value
Mean age at baseline	75.9	75.0	0.54	Included due to significant treatment effect for subgroup age <75 years but not age ≥75 years in Dombret 2015
Sex, male	70.7%	58.4%	0.41	Included due to significant treatment effect for females but not males in Dombret 2015, prognostic in the literature, large imbalance between trials
AML type, de novo	48.3%	82.0%	0.52	Excluded for lack of significance in GLAS + LDAC versus LDAC IPD regression and no subgroup analysis in Dombret 2015
Bone marrow blasts >50%	47.9%	75.4%	0.52	Excluded for lack of significance in GLAS + LDAC versus LDAC IPD regression and lack of significance in Dombret 2015
ECOG PS 0 or 1 versus 2	49.1%	77.4%	0.91	Excluded for lack of significance in GLAS + LDAC versus LDAC IPD regression and no subgroup analysis in Dombret 2015
Cytogenetic risk: poor versus good/intermediate	39.6%	34.8%	-	Included due to being a stratification factor in both trial protocols
Median baseline hemoglobin level (g/dL)	9.2	9.4	0.59	Excluded for lack of significance in GLAS + LDAC versus LDAC IPD regression and no subgroup analysis in Dombret 2015

Abbreviations: AML, acute myeloid leukemia; AZA, azacitidine; DSU, Decision Support Unit; ECOG PS, Eastern Cooperative Oncology Group performance status; GLAS, glasdegib; IPD, individual patient-level data; LDAC, low-dose cytarabine.

Figure 3 Overlay of Kaplan-Meier with exponential parametrization adjusting trial IPD (A) AZA and (B) DEC populations.

Notes: In Figure 3A (AZA) and 3B (DEC), the gray (KM) and both blue (exponential) curves represent OS in the LDAC alone treatment arm. The orange and green lines estimate survival time in the GLAS + LDAC arm. The solid curves apply the average covariate values from the IPD population, while the dashed curves model the mean covariates from the comparator trials (AZA or DEC).
Abbreviations: AZA, azacitidine; DEC, decitabine; GLAS, glasdegib; K-M, Kaplan-Meier; LDAC, low-dose cytarabine; IPD, individual patient data; OS, overall survival.

Figure 4 Overlay of Kaplan-Meier with Weibull parametrization for the weighted STC approach (A) AZA and (B) DEC populations.

Notes: In Figure 4A (AZA) and 4B (DEC), the gray (KM) and both blue (Weibull) curves represent OS in the LDAC alone treatment arm. The orange and green lines estimate survival in the GLAS + LDAC arm. The solid curves apply the average covariate values from the IPD population, while the dashed curves model the mean covariates from the comparator trials (AZA or DEC).
Abbreviations: AZA, azacitidine; DEC, decitabine; GLAS, glasdegib; K-M, Kaplan-Meier; LDAC, low-dose cytarabine; IPD, individual patient data; OS, overall survival; STC, simulated treatment comparison.

Table 3 ITC Cox and STC exponential model results: AZA comparison, DSU guidance

Treatments Compared: model	GLAS + LDAC vs LDAC		AZA vs LDAC (published)		GLAS + LDAC vs AZA
	HR	95% CI	HR	95% CI	HR	95% CI
GLAS + LDAC vs AZA: Cox unadjusted (standard ITC)*	0.463	0.299, 0.717	0.900	0.700, 1.160	0.514	0.310, 0.852
GLAS + LDAC vs AZA: Cox full (multivariate ITC)**	0.418	0.224, 0.779	0.900	0.700, 1.160	0.464	0.237, 0.910
GLAS  +  LDAC vs AZA: stepwise exponential (STC)	0.382	0.217, 0.673	0.900	0.700, 1.160	0.424	0.228, 0.789
GLAS + LDAC vs AZA: full exponential (STC)	0.401	0.219, 0.736	0.900	0.700, 1.160	0.446	0.231, 0.860

Notes: *This row is equivalent to performing standard (unadjusted) ITC comparing GLAS + LDAC to AZA. **This row performs a covariate-adjusted ITC. Bolded values in GLAS + LDAC vs AZA column are meant to highlight results of ITC and STC analysis.

Abbrevations: AZA, azacitidine; GLAS, glasdegib; ITC, indirect treatment comparison; LDAC, low-dose cytarabine; STC, simulated treatment comparison; DSU, Decision Support Unit; HR, hazard ratio.

Table 4 ITC Cox and STC exponential model results: AZA comparison, weighted STC approach

Treatments Compared: Model	GLAS + LDAC vs LDAC		AZA vs LDAC (published)		GLAS + LDAC vs AZA
	HR	95% CI	HR	95% CI	HR	95% CI
GLAS + LDAC vs AZA: Cox unadjusted (standard ITC)*	0.463	0.299, 0.717	0.900	0.700, 1.160	0.514	0.310, 0.852
GLAS + LDAC vs AZA: Cox full (multivariate ITC)**	0.425	0.227, 0.797	0.900	0.700, 1.160	0.473	0.240, 0.930
GLAS + LDAC vs AZA: stepwise Weibull (STC)	0.371	0.203, 0.677	0.900	0.700, 1.160	0.412	0.214, 0.791
GLAS + LDAC vs AZA: full exponential (STC)	0.396	0.216, 0.725	0.900	0.700, 1.160	0.440	0.228, 0.848

Notes: *This row is equivalent to performing standard (unadjusted) ITC comparing GLAS + LDAC to AZA. **This row performs a covariate-adjusted ITC with propensity-weighted GLAS + LDAC vs LDAC data. Bolded values in GLAS + LDAC vs AZA column are meant to highlight results of ITC and STC analysis.

Abbreviations: AZA, azacitidine; GLAS, glasdegib; HR, hazard ratio; ITC, indirect treatment comparison; LDAC, low-dose cytarabine; STC, simulated treatment comparison.

Figure 5 Forest plots of exponential and Cox model estimates for (A) GLAS + LDAC versus AZA and (B) GLAS + LDAC versus DEC, DSU guidance.

Notes: The forest plots (95% confidence intervals) demonstrate GLAS + LDAC superiority vs (A) AZA and (B) DEC, and provide a simple visualization of the comparable HR results among each set of models. The x-axis is presented on the log scale.
Abbreviations: AZA, azacitidine; DEC, decitabine; GLAS, glasdegib; ITC, indirect treatment comparison; LDAC, low-dose cytarabine; STC, simulated treatment comparison.

Figure 6 Forest plots of exponential and Cox model estimates for (A) GLAS + LDAC versus AZA and (B) GLAS + LDAC versus DEC, weighted STC approach.

Notes: The forest plots (95% confidence intervals) demonstrate GLAS + LDAC superiority vs (A) AZA and (B) DEC, and provide a simple visualization of the comparable HR results among each set of models. The x-axis is presented on the log scale.
Abbreviations: AZA, azacitidine; DEC, decitabine; GLAS, glasdegib; HR, hazard ratio; ITC, indirect treatment comparison; LDAC, low-dose cytarabine; STC, simulated treatment comparison.

Table 5 Inclusion of covariates, GLAS + LDAC vs DEC

Full Model Results			Statistical Evidence	Justification for Inclusion in Stepwise Models
Included Baseline Characteristics	GLAS + LDAC (IPD)	DEC (Kantarjian 2012)	GLAS IPD Cox p-value
Mean age at baseline	75.9	73.0	0.54	Included due to significant treatment effect for only subgroup age ≥ 75 years in Kantarjian 2012, potentially prognostic as advised by clinical expertise
Sex, male	70.7%	58.%	0.41	Excluded for lack of subgroup analysis in Kantarjian 2012
AML type, de novo	48.3%	64.6%	0.52	Significant treatment effect for de novo but not secondary AML in Kantarjian 2012, large imbalance between trials
Bone marrow blasts >50%	47.9%	42.7%	0.52	Included due to significant treatment effect for subgroup >30% in Kantarjian 2012, large imbalance between trials
ECOG PS 0 or 1 versus 2	49.1%	76.2%	0.91	Included due to significant treatment effect for subgroup ECOG =2 in Kantarjian 2012, large imbalance between trials
Cytogenetic risk: poor versus good/intermediate	39.6%	36.3%	-	Included due to being a stratification factor in both trial protocols
Median hemoglobin at baseline	9.2	9.3	0.59	Excluded for lack of significance in GLAS + LDAC versus LDAC IPD regression and no subgroup analysis in Kantarjian 2012

Abbreviations: AML, acute myeloid leukemia; DEC, decitabine; DSU, Decision Support Unit; ECOG PS, Eastern Cooperative Oncology Group performance status; GLAS, glasdegib; LDAC, low-dose cytarabine.

Table 6 ITC Cox and STC exponential model results: DEC comparison, DSU guidance

Treatments Compared: Model	GLAS + LDAC vs LDAC		DEC vs LDAC (published)		GLAS + LDAC vs DEC
	HR	95% CI	HR	95% CI	HR	95% CI
GLAS + LDAC vs DEC: Cox unadjusted (standard ITC)*	0.463	0.299, 0.717	0.820	0.680, 0.990	0.565	0.351, 0.909
GLAS + LDAC vs DEC: Cox full (multivariate ITC)**	0.418	0.224, 0.779	0.820	0.680, 0.990	0.510	0.266, 0.977
GLAS + LDAC vs DEC: stepwise exponential (STC)	0.414	0.227, 0.757	0.820	0.680, 0.990	0.505	0.269, 0.949
GLAS + LDAC vs DEC: STC full exponential (STC)	0.401	0.219, 0.736	0.820	0.680, 0.990	0.490	0.259, 0.924

Notes: *This row is equivalent to performing standard (unadjusted) ITC comparing GLAS + LDAC to DEC. **This row performs a covariate-adjusted ITC. Bolded values in GLAS + LDAC vs AZA column are meant to highlight results of ITC and STC analysis.

Abbreviations: DEC, decitabine; DSU, Decision Support Unit; HR, hazard ratio; GLAS, glasdegib; ITC, indirect treatment comparison; LDAC, low-dose cytarabine; STC, simulated treatment comparison.

Table 7 ITC Cox and STC exponential model results: DEC comparison, weighted STC approach

Treatments Compared: Model	GLAS + LDAC vs LDAC		DEC vs LDAC (published)		GLAS + LDAC vs DEC
	HR	95% CI	HR	95% CI	HR	95% CI
GLAS + LDAC vs DEC: Cox unadjusted (standard ITC)*	0.463	0.299, 0.717	0.820	0.680, 0.990	0.565	0.351, 0.909
GLAS + LDAC vs DEC: Cox full (multivariate ITC)**	0.422	0.225, 0.792	0.820	0.680, 0.990	0.515	0.267, 0.992
GLAS + LDAC vs DEC: stepwise Weibull (STC)	0.397	0.204, 0.772	0.820	0.680, 0.990	0.484	0.242, 0.967
GLAS + LDAC vs DEC: STC full exponential (STC)	0.395	0.215, 0.725	0.820	0.680, 0.990	0.482	0.255, 0.909

Notes: *This row is equivalent to performing standard (unadjusted) ITC comparing GLAS + LDAC to DEC. **This row performs a covariate-adjusted ITC with propensity-weighted GLAS + LDAC vs LDAC data. Bolded values in GLAS + LDAC vs AZA column are meant to highlight results of ITC and STC analysis.

Abbreviations: DEC, decitabine; GLAS, glasdegib; HR, hazard ratio; ITC, indirect treatment comparison; LDAC, low-dose cytarabine; STC, simulated treatment comparison.

Spriggs D, Griffin J, Wisch J, Kufe D. Clinical pharmacology of low-dose cytosine arabinoside. Blood. 1985;65(5):1087–1089.3995167

 PubMed Web of Science ®Google Scholar