Table 1 Clinicopathologic and trial characteristics.

From: Prostate cancer therapy personalization via multi-modal deep learning on randomized phase III clinical trials

 

RTOG-9202

RTOG-9408

RTOG-9413

RTOG-9910

RTOG-0126

Combined

Number of Patients

1180

1719

695

976

1084

5654

White

1004

1312

481

769

937

4503

Hispanic

18

48

23

23

0

112

African American

147

334

173

166

112

932

Asian

4

12

1

11

12

40

Other Race

1

10

11

6

9

37

Unknown Race

6

3

6

1

14

30

Number of Pathology Slides

3188

5472

2104

3075

2365

16204

Number of Clinical Variables

53

69

71

60

62

Therapy Randomization

RTS vs. RTL

RT vs. RTS

RTS 2 × 2a

RTS vs. RTM

RT vs. RT + 

Patient Risk Groups

Inter. | High

Low | Inter. | High

Inter. | High

Inter. | High

Inter.

Low | Inter. | High

Primary Endpoint

Disease-free Survival

Overall Survival

Progression-free Survival

Prostate Cancer-specific Mortality

Overall Survival

Median Follow-up for Censored Patients (Years)

17.4

15.1

13.7

9.3

13.2

11.4

No. Patients Died

944

1154

504

297

505

3404

Trial Accrual Dates

1992–1995

1994–2001

1995–1999

2000–2004

2002–2008

1992–2008

  1. The column ‘combined’ shows the characteristics of the final dataset with all five trials used for training and validation. aRTOG-9413 randomized patients in a 2 × 2 fashion testing the effect of timing of ADT (before and during RT vs. starting after RT) and field size (prostate only vs. full pelvic RT). New acronyms: radiotherapy plus short/medium/long-term hormone therapy (RTS/RTM/RTL).