An integrated prediction model of recurrence in endometrial endometrioid cancers

Figures & data

Figure 1 Flow chart of patients included in the UIHC endometrial cancer study cohort. There were 126 patients with endometrial cancer, endometrioid type. 62 had sufficient quantity and quality of purified RNA for RNA sequencing.

Abbreviation: CHA, complex endometrial hyperplasia with atypia; UIHC, University of Iowa Hospitals and Clinics.

Table 1 Clinical and pathological characteristics of the UIHC cohort of patients included in this study

		Recurred (N=16)	Not recurred (N=109)	p-value
Preoperative characteristics	Age (mean)	65	61	0.093
	BMI (mean)	31.9	36.4	0.114
	Charlson Morbidity Index	5.4	5	0.271
	Grade			0.122
	1	2	42
	2	9	39
	3	5	25
	Level of risk			0.003
	Low	1	69
High	15	40
Postoperative characteristics	Invasion (mean)	67	34	<0.001
	2009 FIGO Stage			<0.001
	I	4	89
	II	3	4
	III	8	11
	IV	1	5
	Lymph nodes (positive)	6 (50%)	6 (7%)	<0.001
	Peritoneal Cytology (positive)	4 (31%)	8 (8%)	0.012
	Lymphovascular involvement	11 (73%)	19 (20%)	<0.001
	ER (positive)	9 (82%)	60 (86%)	0.706
	PR (positive)	7 (64%)	61 (87%)	0.035
	Type of surgery (MI)	2 (17%)	8 (13%)	0.502
	Postoperative complications	5 (21%)	23 (23%)	0.908
	LOS (days)	4.9	4.3	0.723
	Adjuvant treatment (yes)	10 (63%)	37 (34%)	0.017
Adjuvant radiation (yes)	5 (31%)	26 (19%)	0.264
Outcomes	Overall survival (5 years)	17%	90%	<0.001
Death due to disease	11 (69%)	0 (0%)	<0.001

Abbreviations: UIHC, University of Iowa Hospitals and Clinics; BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics; ER, estrogen receptor positive; PR, progesterone receptor positive; MI, minimally invasive; LOS, length of stay.

Figure 2 Representation of differential gene expression (A), miRNA expression (B), somatic mutation (C), and copy number variation (D) by recurrence in the UIHC patient cohort (n=62 patients total; n=8 recurrence and n=54 no recurrence).

Abbreviation: CNV, copy number variation; UIHC, University of Iowa Hospitals and Clinics.

Table 2 Prediction models for recurrence incorporating clinical, pathological and molecular data, as well as external replication of prediction models in TCGA

Prediction models including 1 category of data
Model	Data included	AUC	95% CI
UIHC model	Clinical	0.90	0.87, 0.92
TCGA model	Clinical	0.66	0.61, 0.72
UIHC model	mRNA	0.74	0.62, 0.87
TCGA model	mRNA	0.56	0.50, 0.62
UIHC model	miRNA	0.81	0.77, 0.86
TCGA model	miRNA	0.64	0.60, 0.69
UIHC model	Mutations	0.60	0.49, 0.70
TCGA model	Mutations	0.54	0.52, 0.56
UIHC model	CNV	0.68	0.63, 0.74
TCGA model	CNV	0.70	0.65, 0.75
Prediction models including 2 categories of data
Model	Data included: clinical+	AUC	95% CI
UIHC model	mRNA	0.99	0.98, 1.0
TCGA model	mRNA	0.60	0.56, 0.63
UIHC model	miRNA	0.92	0.88, 096
TCGA model	miRNA	0.66	0.62,0.70
UIHC model	Mutations	0.96	0.94, 0.98
TCGA model	Mutations	0.62	0.57, 0.66
UIHC model	CNV	1.0	1.0, 1.0
TCGA model	CNV	0.72	0.65, 0.79
Prediction models including 3 categories of data
Model	Data included: clinical +	AUC	95% CI
UIHC model	mRNA+miRNA	0.84	0.81, 0.88
TCGA model	mRNA+miRNA	0.61	0.55, 0.66
UIHC model	Mutations+CNV	0.94	0.91, 0.98
TCGA model	Mutations+CNV	0.71	0.64, 0.78
UIHC model	mRNA+mutations	0.84	0.76, 0.92
TCGA model	mRNA+mutations	0.63	0.58, 0.66
UIHC model	Mutations+miRNA	0.89	0.84, 0.95
TCGA model	Mutations+miRNA	0.62	0.56, 0.67
UIHC model	CNV+miRNA	0.91	0.86, 0.96
TCGA model	CNV+miRNA	0.71	0.64, 0.78
UIHC model	mRNA+CNV	0.92	0.85, 0.98
TCGA model	mRNA+CNV	0.70	0.63, 0.77
Prediction models including 3 categories of data
Model	Data included: clinical+	AUC	95% CI
UIHC model	mRNA+mutations+CNV	0.85	0.77, 0.92
TCGA model	mRNA+mutations+CNV	0.69	0.63, 0.75
UIHC model	mRNA+miRNA+CNV	0.84	0.77, 0.91
TCGA model	mRNA+miRNA+CNV	0.70	0.63, 0.77
UIHC model	miRNA+CNV+mutations	0.92	0.89, 0.95
TCGA model	miRNA+CNV+mutations	0.70	0.63, 0.77
UIHC model	mRNA+miRNA+Mutations	0.89	0.84, 0.95
TCGA model	mRNA+miRNA+Mutations	0.62	0.58, 0.67
Prediction models including 5 categories of data
Model	Data included: clinical+	AUC	95% CI
UIHC model	mRNA+miRNA+ Mutations+CNV	0.89	0.84, 0.95
TCGA model	mRNA+miRNA + Mutations+CNV	0.69	0.62, 0.75

Note: For the replication in TCGA we included the same variables as those in the UIHC analysis. There were few variables that were selected for analysis in UIHC data that were not found in TCGA data: 2 mRNA and 2 somatic mutations. Therefore, in models with multiple categories of data (2 or more) we included the variables resulting from the UIHC best prediction model with one category of data: 1 clinical variable (risk level); 21 mRNAs (19 for TCGA); 15 miRNAs; 22 somatic mutations (20 for TCGA); and 43 copies of genes.

Abbreviation: AUC, area under the curve; UIHC, University of Iowa Hospitals and Clinics; CNV, copy number variation; TCGA, the Cancer Genome Atlas.

Table 3 Values for each individual variable used to construct the prediction model score

Clinical variables
Level of risk	Low risk=1	High risk=2
Molecular variables
	miRNA	Log2 transformed and normalized miRNA expression for:
		MIR217
		MIR224
		MIR301B
		MIR3196
		MIR3974
		MIR4285
		MIR4420
		MIR4643
		MIR4788
	MIR6811
	Copy number variation	Log2 transformed and normalized copy number counts for segments:
			Genes included
		chr1:1874618-1925130	KIAA1751
		chr1:19806824-19807719	LOC644068
		chr2:167857933-168123083	LOC643496
		chr2:179972700-179973847	LOC644776
		chr2:201343640-201367191	AOX2
		chr2:238893821-238972536	TRAF3IP1
		chr3:48173672-48204805	CDC25A
		chr4:189153919-189163193	ZFP42
		chr5:56649667-56650212	LOC402217
		chr6:52950710-52968099	GSTA4
		chr6:52968377-52968708	RN7SK
		chr7:35638794-35701254	FLJ22313
		chr7:38166835-38169796	LOC646955
		chr7:100625730-100631022	MOGAT3
		chr7:100631763-100634385	LOC646409
		chr7:100635978-100647731	PLOD3
		chr9:467739-493664	LOC645577
		chr11:2422797-2826916	KCNQ1
		chr11:76171933-76186846	LRRC54
		chr11:85829798-86061075	ME3
		chr12:31792077-31797910	LOC645636
chr12:31798855-31799506	LOC144383
		chr12:31835386-31836442	LOC440093
		chr12:104429458-104438355	LOC644452
		chr13:48720099-48765619	CDADC1
		chr15:72620637-72677525	ARID3B
		chr15:72687766-72709511	CLK3
		chr19:40921991-40925191	U2AF1L4
		chr19:40925272-40928145	U2AF1L3
		chr19:40928334-40929743	PSENEN
		chr20:45271788-45418881	PRKCBP1
		chr22:21094316-21094614	IGLV1-40
		chr22:21104714-21106124	ASH2LP1
		chr22:25209846-25217899	LOC402055
		chr22:35290050-35428849	CACNG2

Table 4 Validation of the prediction model of recurrence in TCGA

	Model with clinical/copy number variation (CNV)		Model with clinical/miRNA		Model with clinical/miRNA/CNV
Recurrence probability scale*	Cutoff=0.501		Cutoff=0.500		Cutoff=0.553
	Value	95% CI	Value	95% CI	Value	95% CI
Sensitivity	55%	0.48, 0.61	33%	0.18, 0.47	32%	0.13, 0.52
Specificity	81%	0.65, 0.94	81%	0.77, 0.85	80%	0.75, 0.85
Positive predictive value	16%	0.07, 0.25	19%	0.12, 0.25	18%	0.08, 0.26
Negative predictive value	89%	0.87, 0.91	89%	0.88, 0.91	90%	0.87, 0.92
Accuracy	74%	0.72, 0.77	75%	0.73, 0.77	74%	0.72, 0.77

Note: *Recurrence probability scale: 1/(exp(-score)+1), where score is the resulting value of the prediction model in logit scale. As detailed in methods, the threshold was selected for specificity around 80% and highest sensitivity and negative predictive value. The goal was to create models that would rule out at least 80 of non-recurrent patients but still capturing most of patients with recurrence.

Figure S1 Overall survival by recurrence status.

Table S1 TCGA Patient clinical and pathological characteristics (N=394). Univariate analysis with Cox proportional Hazard ratio was used to assess differences between both groups

		Recurred (N=49)	Not recurred (N=345)	p-value
Preoperative characteristics	Age (mean)	63	62	0.618
	BMI (mean)	33.2	33.1	0.742
	Grade			0.001
	1	4	93
	2	14	99
	3	31	153
	Level of risk			0.007
Low	17	189
	High	32	156
Postoperative characteristics	Myometrial invasion			0.515
	<50%	22	238
	>50%	3	14
	2009 FIGO Stage			<0.001
	I	25	252
	II	2	31
	III	16	55
	IV	6	7
	Lymph nodes (positive)	14 (33%)	27 (10%)	<0.001
Peritoneal Cytology (positive)	8 (18%)	20 (8%)	0.024

Abbreviations: BMI, body mass index; FIGO, International Federation of Gynecology and Obstetrics; TCGA, the Cancer Genome Atlas.