Glioblastoma Multiforme: Correlation between molecular cancer subtypes and selected clinical features

Maintained by TCGA GDAC Team (Broad Institute/Dana-Farber Cancer Institute/Harvard Medical School)

Overview

Introduction

This pipeline computes the correlation between cancer subtypes identified by different molecular patterns and selected clinical features.

Summary

Testing the association between subtypes identified by 5 different clustering approaches and 6 clinical features across 527 patients, 7 significant findings detected with P value < 0.05.

CNMF clustering analysis on array-based mRNA expression data identified 3 subtypes that correlate to 'AGE'.
Consensus hierarchical clustering analysis on array-based mRNA expression data identified 3 subtypes that correlate to 'AGE'.
CNMF clustering analysis on array-based miR expression data identified 4 subtypes that correlate to 'Time to Death'.
Consensus hierarchical clustering analysis on array-based miR expression data identified 3 subtypes that correlate to 'Time to Death' and 'AGE'.
3 subtypes identified in current cancer cohort by 'METHLYATION CNMF'. These subtypes correlate to 'Time to Death' and 'AGE'.

Results

Overview of the results

Table 1. Get Full Table Overview of the association between subtypes identified by 5 different clustering approaches and 6 clinical features. Shown in the table are P values from statistical tests. Thresholded by P value < 0.05, 7 significant findings detected.


Clinical Features	Statistical Tests	mRNA CNMF subtypes	mRNA cHierClus subtypes	miR CNMF subtypes	miR cHierClus subtypes	METHLYATION CNMF
Time to Death	logrank test	0.157	0.056	0.000614	0.00455	0.00142
AGE	ANOVA	0.0257	0.0276	0.101	0.000872	5.2e-08
GENDER	Fisher's exact test	0.497	0.563	0.508	0.119	0.849
KARNOFSKY PERFORMANCE SCORE	ANOVA	0.839	0.487	0.943	0.785	0.223
RADIATIONS RADIATION REGIMENINDICATION	Fisher's exact test	0.259	0.122	0.638	0.758	0.176
NEOADJUVANT THERAPY	Fisher's exact test	0.821	0.633	0.906	0.993	0.484

Clustering Approach #1: 'mRNA CNMF subtypes'

Table S1. Get Full Table Description of clustering approach #1: 'mRNA CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	177	172	170

'mRNA CNMF subtypes' versus 'Time to Death'

P value = 0.157 (logrank test)

Table S2. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	519	403	0.1 - 127.6 (9.9)
subtype1	177	145	0.2 - 127.6 (10.0)
subtype2	172	129	0.2 - 108.8 (9.2)
subtype3	170	129	0.1 - 92.6 (10.7)

Figure S1. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'mRNA CNMF subtypes' versus 'AGE'

P value = 0.0257 (ANOVA)

Table S3. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	519	57.7 (14.5)
subtype1	177	57.3 (12.8)
subtype2	172	55.8 (16.4)
subtype3	170	60.0 (13.7)

Figure S2. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #2: 'AGE'

'mRNA CNMF subtypes' versus 'GENDER'

P value = 0.497 (Fisher's exact test)

Table S4. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	204	315
subtype1	70	107
subtype2	62	110
subtype3	72	98

Figure S3. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'mRNA CNMF subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.839 (ANOVA)

Table S5. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	389	77.1 (14.4)
subtype1	137	77.5 (15.0)
subtype2	126	77.3 (13.0)
subtype3	126	76.5 (15.0)

Figure S4. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'mRNA CNMF subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.259 (Fisher's exact test)

Table S6. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	348	171
subtype1	119	58
subtype2	108	64
subtype3	121	49

Figure S5. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'mRNA CNMF subtypes' versus 'NEOADJUVANT.THERAPY'

P value = 0.821 (Fisher's exact test)

Table S7. Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

nPatients	NO	YES
ALL	270	249
subtype1	94	83
subtype2	86	86
subtype3	90	80

Figure S6. Get High-res Image Clustering Approach #1: 'mRNA CNMF subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

Clustering Approach #2: 'mRNA cHierClus subtypes'

Table S8. Get Full Table Description of clustering approach #2: 'mRNA cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	223	127	169

'mRNA cHierClus subtypes' versus 'Time to Death'

P value = 0.056 (logrank test)

Table S9. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	519	403	0.1 - 127.6 (9.9)
subtype1	223	182	0.1 - 90.6 (10.4)
subtype2	127	94	0.1 - 92.6 (9.8)
subtype3	169	127	0.2 - 127.6 (9.4)

Figure S7. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'mRNA cHierClus subtypes' versus 'AGE'

P value = 0.0276 (ANOVA)

Table S10. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	519	57.7 (14.5)
subtype1	223	57.7 (13.4)
subtype2	127	60.3 (13.8)
subtype3	169	55.7 (16.0)

Figure S8. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'mRNA cHierClus subtypes' versus 'GENDER'

P value = 0.563 (Fisher's exact test)

Table S11. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	204	315
subtype1	90	133
subtype2	53	74
subtype3	61	108

Figure S9. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'mRNA cHierClus subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.487 (ANOVA)

Table S12. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	389	77.1 (14.4)
subtype1	165	77.6 (14.4)
subtype2	98	75.6 (15.4)
subtype3	126	77.6 (13.5)

Figure S10. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'mRNA cHierClus subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.122 (Fisher's exact test)

Table S13. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	348	171
subtype1	156	67
subtype2	89	38
subtype3	103	66

Figure S11. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'mRNA cHierClus subtypes' versus 'NEOADJUVANT.THERAPY'

P value = 0.633 (Fisher's exact test)

Table S14. Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

nPatients	NO	YES
ALL	270	249
subtype1	118	105
subtype2	69	58
subtype3	83	86

Figure S12. Get High-res Image Clustering Approach #2: 'mRNA cHierClus subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

Clustering Approach #3: 'miR CNMF subtypes'

Table S15. Get Full Table Description of clustering approach #3: 'miR CNMF subtypes'

Cluster Labels	1	2	3	4
Number of samples	144	159	74	105

'miR CNMF subtypes' versus 'Time to Death'

P value = 0.000614 (logrank test)

Table S16. Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	482	379	0.1 - 127.6 (10.3)
subtype1	144	116	0.1 - 51.3 (10.6)
subtype2	159	124	0.1 - 127.6 (10.6)
subtype3	74	57	0.1 - 53.8 (8.4)
subtype4	105	82	0.1 - 92.6 (10.8)

Figure S13. Get High-res Image Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'miR CNMF subtypes' versus 'AGE'

P value = 0.101 (ANOVA)

Table S17. Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	482	57.5 (14.6)
subtype1	144	59.7 (11.4)
subtype2	159	55.5 (17.0)
subtype3	74	57.9 (15.3)
subtype4	105	57.4 (13.7)

Figure S14. Get High-res Image Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #2: 'AGE'

'miR CNMF subtypes' versus 'GENDER'

P value = 0.508 (Fisher's exact test)

Table S18. Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	186	296
subtype1	56	88
subtype2	68	91
subtype3	26	48
subtype4	36	69

Figure S15. Get High-res Image Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #3: 'GENDER'

'miR CNMF subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.943 (ANOVA)

Table S19. Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	368	77.6 (14.1)
subtype1	112	77.6 (14.4)
subtype2	114	77.5 (14.1)
subtype3	62	76.9 (14.4)
subtype4	80	78.4 (13.8)

Figure S16. Get High-res Image Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'miR CNMF subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.638 (Fisher's exact test)

Table S20. Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	327	155
subtype1	102	42
subtype2	108	51
subtype3	46	28
subtype4	71	34

Figure S17. Get High-res Image Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'miR CNMF subtypes' versus 'NEOADJUVANT.THERAPY'

P value = 0.906 (Fisher's exact test)

Table S21. Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

nPatients	NO	YES
ALL	248	234
subtype1	75	69
subtype2	78	81
subtype3	39	35
subtype4	56	49

Figure S18. Get High-res Image Clustering Approach #3: 'miR CNMF subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

Clustering Approach #4: 'miR cHierClus subtypes'

Table S22. Get Full Table Description of clustering approach #4: 'miR cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	170	180	132

'miR cHierClus subtypes' versus 'Time to Death'

P value = 0.00455 (logrank test)

Table S23. Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	482	379	0.1 - 127.6 (10.3)
subtype1	170	137	0.1 - 92.6 (9.8)
subtype2	180	145	0.1 - 127.6 (10.0)
subtype3	132	97	0.1 - 108.8 (10.7)

Figure S19. Get High-res Image Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'miR cHierClus subtypes' versus 'AGE'

P value = 0.000872 (ANOVA)

Table S24. Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	482	57.5 (14.6)
subtype1	170	59.2 (12.6)
subtype2	180	58.9 (13.3)
subtype3	132	53.5 (17.6)

Figure S20. Get High-res Image Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'miR cHierClus subtypes' versus 'GENDER'

P value = 0.119 (Fisher's exact test)

Table S25. Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	186	296
subtype1	58	112
subtype2	80	100
subtype3	48	84

Figure S21. Get High-res Image Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #3: 'GENDER'

'miR cHierClus subtypes' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.785 (ANOVA)

Table S26. Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	368	77.6 (14.1)
subtype1	129	78.3 (13.5)
subtype2	136	77.1 (15.5)
subtype3	103	77.4 (13.1)

Figure S22. Get High-res Image Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'miR cHierClus subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.758 (Fisher's exact test)

Table S27. Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	327	155
subtype1	113	57
subtype2	121	59
subtype3	93	39

Figure S23. Get High-res Image Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'miR cHierClus subtypes' versus 'NEOADJUVANT.THERAPY'

P value = 0.993 (Fisher's exact test)

Table S28. Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

nPatients	NO	YES
ALL	248	234
subtype1	88	82
subtype2	92	88
subtype3	68	64

Figure S24. Get High-res Image Clustering Approach #4: 'miR cHierClus subtypes' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

Clustering Approach #5: 'METHLYATION CNMF'

Table S29. Get Full Table Description of clustering approach #5: 'METHLYATION CNMF'

Cluster Labels	1	2	3
Number of samples	79	123	78

'METHLYATION CNMF' versus 'Time to Death'

P value = 0.00142 (logrank test)

Table S30. Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Month
ALL	280	208	0.1 - 127.6 (10.0)
subtype1	79	57	0.1 - 92.6 (10.0)
subtype2	123	95	0.1 - 77.6 (9.4)
subtype3	78	56	0.2 - 127.6 (11.7)

Figure S25. Get High-res Image Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

'METHLYATION CNMF' versus 'AGE'

P value = 5.2e-08 (ANOVA)

Table S31. Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	280	57.5 (14.9)
subtype1	79	57.0 (11.9)
subtype2	123	62.4 (12.6)
subtype3	78	50.3 (17.7)

Figure S26. Get High-res Image Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #2: 'AGE'

'METHLYATION CNMF' versus 'GENDER'

P value = 0.849 (Fisher's exact test)

Table S32. Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #3: 'GENDER'

nPatients	FEMALE	MALE
ALL	114	166
subtype1	34	45
subtype2	50	73
subtype3	30	48

Figure S27. Get High-res Image Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #3: 'GENDER'

'METHLYATION CNMF' versus 'KARNOFSKY.PERFORMANCE.SCORE'

P value = 0.223 (ANOVA)

Table S33. Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

	nPatients	Mean (Std.Dev)
ALL	213	75.4 (15.0)
subtype1	61	77.0 (17.1)
subtype2	93	73.3 (14.8)
subtype3	59	76.8 (12.5)

Figure S28. Get High-res Image Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #4: 'KARNOFSKY.PERFORMANCE.SCORE'

'METHLYATION CNMF' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.176 (Fisher's exact test)

Table S34. Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	206	74
subtype1	63	16
subtype2	84	39
subtype3	59	19

Figure S29. Get High-res Image Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #5: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'METHLYATION CNMF' versus 'NEOADJUVANT.THERAPY'

P value = 0.484 (Fisher's exact test)

Table S35. Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

nPatients	NO	YES
ALL	164	116
subtype1	49	30
subtype2	67	56
subtype3	48	30

Figure S30. Get High-res Image Clustering Approach #5: 'METHLYATION CNMF' versus Clinical Feature #6: 'NEOADJUVANT.THERAPY'

Methods & Data

Input

Cluster data file = GBM.mergedcluster.txt
Clinical data file = GBM.clin.merged.picked.txt
Number of patients = 527
Number of clustering approaches = 5
Number of selected clinical features = 6
Exclude small clusters that include fewer than K patients, K = 3

Clustering approaches

CNMF clustering

consensus non-negative matrix factorization clustering approach (Brunet et al. 2004)

Consensus hierarchical clustering

Resampling-based clustering method (Monti et al. 2003)

Survival analysis

For survival clinical features, the Kaplan-Meier survival curves of tumors with and without gene mutations were plotted and the statistical significance P values were estimated by logrank test (Bland and Altman 2004) using the 'survdiff' function in R

ANOVA analysis

For continuous numerical clinical features, one-way analysis of variance (Howell 2002) was applied to compare the clinical values between tumor subtypes using 'anova' function in R

Fisher's exact test

For binary clinical features, two-tailed Fisher's exact tests (Fisher 1922) were used to estimate the P values using the 'fisher.test' function in R

Download Results

This is an experimental feature. The full results of the analysis summarized in this report can be downloaded from the TCGA Data Coordination Center.

References

[1] Brunet et al., Metagenes and molecular pattern discovery using matrix factorization, PNAS 101(12):4164-9 (2004)

[2] Monti et al., Consensus Clustering: A Resampling-Based Method for Class Discovery and Visualization of Gene Expression Microarray Data, Machine Learning 52(1):91-118 (2003)

[3] Bland and Altman, Statistics notes: The logrank test, BMJ 328(7447):1073 (2004)

[4] Howell, D, Statistical Methods for Psychology. (5th ed.), Duxbury Press:324-5 (2002)

[5] Fisher, R.A., On the interpretation of chi-square from contingency tables, and the calculation of P, Journal of the Royal Statistical Society 85(1):87-94 (1922)

Made with Nozzle