Correlation between aggregated molecular cancer subtypes and selected clinical features

Head and Neck Squamous Cell Carcinoma (Primary solid tumor)

15 July 2014 | analyses__2014_07_15

Maintainer Information

Citation Information

Maintained by TCGA GDAC Team (Broad Institute/MD Anderson Cancer Center/Harvard Medical School)

Cite as Broad Institute TCGA Genome Data Analysis Center (2014): Correlation between aggregated molecular cancer subtypes and selected clinical features. Broad Institute of MIT and Harvard. doi:10.7908/C16T0KC5

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 10 different clustering approaches and 13 clinical features across 423 patients, 11 significant findings detected with P value < 0.05 and Q value < 0.25.

4 subtypes identified in current cancer cohort by 'Copy Number Ratio CNMF subtypes'. These subtypes correlate to 'NUMBERPACKYEARSSMOKED'.
7 subtypes identified in current cancer cohort by 'METHLYATION CNMF'. These subtypes correlate to 'AGE', 'GENDER', 'NUMBERPACKYEARSSMOKED', and 'NUMBER.OF.LYMPH.NODES'.
CNMF clustering analysis on RPPA data identified 4 subtypes that do not correlate to any clinical features.
Consensus hierarchical clustering analysis on RPPA data identified 3 subtypes that correlate to 'PATHOLOGY.N.STAGE' and 'NUMBER.OF.LYMPH.NODES'.
CNMF clustering analysis on sequencing-based mRNA expression data identified 3 subtypes that do not correlate to any clinical features.
Consensus hierarchical clustering analysis on sequencing-based mRNA expression data identified 4 subtypes that correlate to 'GENDER' and 'NUMBER.OF.LYMPH.NODES'.
3 subtypes identified in current cancer cohort by 'MIRSEQ CNMF'. These subtypes do not correlate to any clinical features.
3 subtypes identified in current cancer cohort by 'MIRSEQ CHIERARCHICAL'. These subtypes do not correlate to any clinical features.
3 subtypes identified in current cancer cohort by 'MIRseq Mature CNMF subtypes'. These subtypes correlate to 'RADIATIONS.RADIATION.REGIMENINDICATION'.
5 subtypes identified in current cancer cohort by 'MIRseq Mature cHierClus subtypes'. These subtypes correlate to 'NEOPLASM.DISEASESTAGE'.

Results

Overview of the results

Table 1. Get Full Table Overview of the association between subtypes identified by 10 different clustering approaches and 13 clinical features. Shown in the table are P values (Q values). Thresholded by P value < 0.05 and Q value < 0.25, 11 significant findings detected.


Clinical Features	Statistical Tests	Copy Number Ratio CNMF subtypes	METHLYATION CNMF	RPPA CNMF subtypes	RPPA cHierClus subtypes	RNAseq CNMF subtypes	RNAseq cHierClus subtypes	MIRSEQ CNMF	MIRSEQ CHIERARCHICAL	MIRseq Mature CNMF subtypes	MIRseq Mature cHierClus subtypes
Time to Death	logrank test	0.732 (1.00)	0.0801 (1.00)	0.647 (1.00)	0.146 (1.00)	0.259 (1.00)	0.301 (1.00)	0.885 (1.00)	0.00776 (0.823)	0.337 (1.00)	0.178 (1.00)
AGE	Kruskal-Wallis (anova)	0.0613 (1.00)	3.2e-07 (4.03e-05)	0.255 (1.00)	0.0759 (1.00)	0.564 (1.00)	0.00917 (0.962)	0.269 (1.00)	0.0419 (1.00)	0.634 (1.00)	0.807 (1.00)
NEOPLASM DISEASESTAGE	Fisher's exact test	0.673 (1.00)	0.83 (1.00)	0.521 (1.00)	0.0299 (1.00)	0.141 (1.00)	0.474 (1.00)	0.312 (1.00)	0.121 (1.00)	0.34 (1.00)	0.00161 (0.187)
PATHOLOGY T STAGE	Fisher's exact test	0.436 (1.00)	0.158 (1.00)	0.17 (1.00)	0.194 (1.00)	0.226 (1.00)	0.033 (1.00)	0.141 (1.00)	0.00483 (0.536)	0.217 (1.00)	0.00231 (0.266)
PATHOLOGY N STAGE	Fisher's exact test	0.0857 (1.00)	0.198 (1.00)	0.357 (1.00)	3e-05 (0.00369)	0.238 (1.00)	0.00343 (0.388)	0.193 (1.00)	0.0716 (1.00)	0.466 (1.00)	0.0382 (1.00)
PATHOLOGY M STAGE	Fisher's exact test	0.864 (1.00)	0.0468 (1.00)			0.447 (1.00)	0.215 (1.00)	0.16 (1.00)	0.204 (1.00)	0.0166 (1.00)	0.32 (1.00)
GENDER	Fisher's exact test	0.0367 (1.00)	2e-05 (0.00248)	0.00352 (0.394)	0.218 (1.00)	0.0124 (1.00)	0.00103 (0.123)	0.372 (1.00)	0.0239 (1.00)	0.0537 (1.00)	0.424 (1.00)
HISTOLOGICAL TYPE	Fisher's exact test	0.0989 (1.00)	0.0464 (1.00)			0.00257 (0.293)	0.0472 (1.00)	0.135 (1.00)	0.371 (1.00)	0.688 (1.00)	0.179 (1.00)
RADIATIONS RADIATION REGIMENINDICATION	Fisher's exact test	0.198 (1.00)	0.509 (1.00)	0.00683 (0.738)	0.436 (1.00)	0.286 (1.00)	0.0266 (1.00)	0.0555 (1.00)	0.127 (1.00)	0.00128 (0.151)	0.0869 (1.00)
NUMBERPACKYEARSSMOKED	Kruskal-Wallis (anova)	0.000258 (0.031)	0.00134 (0.157)	0.0841 (1.00)	0.746 (1.00)	0.015 (1.00)	0.0442 (1.00)	0.162 (1.00)	0.272 (1.00)	0.0069 (0.739)	0.0392 (1.00)
NUMBER OF LYMPH NODES	Kruskal-Wallis (anova)	0.0123 (1.00)	0.000253 (0.0307)	0.208 (1.00)	4.76e-06 (0.000595)	0.0385 (1.00)	3.29e-05 (0.00401)	0.0152 (1.00)	0.005 (0.55)	0.0614 (1.00)	0.0412 (1.00)
RACE	Fisher's exact test	0.501 (1.00)	0.13 (1.00)	0.112 (1.00)	0.418 (1.00)	0.485 (1.00)	0.0491 (1.00)	0.291 (1.00)	0.019 (1.00)	0.0536 (1.00)	0.00655 (0.714)
ETHNICITY	Fisher's exact test	0.187 (1.00)	0.698 (1.00)	0.74 (1.00)	0.54 (1.00)	0.265 (1.00)	0.984 (1.00)	0.443 (1.00)	0.304 (1.00)	0.721 (1.00)	0.248 (1.00)

Clustering Approach #1: 'Copy Number Ratio CNMF subtypes'

Table S1. Description of clustering approach #1: 'Copy Number Ratio CNMF subtypes'

Cluster Labels	1	2	3	4
Number of samples	149	83	151	34

'Copy Number Ratio CNMF subtypes' versus 'Time to Death'

P value = 0.732 (logrank test), Q value = 1

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

	nPatients	nDeath	Duration Range (Median), Year
ALL	355	98	2.0 - 3837.0 (629.0)
subtype1	120	42	2.0 - 3837.0 (558.5)
subtype2	73	21	6.0 - 3295.0 (613.0)
subtype3	132	27	4.0 - 3835.0 (723.5)
subtype4	30	8	3.0 - 2327.0 (568.0)

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

'Copy Number Ratio CNMF subtypes' versus 'AGE'

P value = 0.0613 (Kruskal-Wallis (anova)), Q value = 1

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

	nPatients	Mean (Std.Dev)
ALL	416	60.9 (12.2)
subtype1	149	59.2 (12.7)
subtype2	83	61.7 (10.1)
subtype3	150	61.1 (12.8)
subtype4	34	65.3 (10.2)

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

'Copy Number Ratio CNMF subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.673 (Fisher's exact test), Q value = 1

Table S4. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	24	63	71	192	9
subtype1	9	18	23	77	5
subtype2	3	11	13	41	1
subtype3	9	28	27	60	2
subtype4	3	6	8	14	1

Figure S3. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.436 (Fisher's exact test), Q value = 1

Table S5. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	40	108	85	131
subtype1	14	36	27	56
subtype2	5	21	21	23
subtype3	17	43	26	43
subtype4	4	8	11	9

Figure S4. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 0.0857 (Fisher's exact test), Q value = 1

Table S6. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	136	56	129	7
subtype1	40	19	59	3
subtype2	24	9	28	1
subtype3	59	21	33	2
subtype4	13	7	9	1

Figure S5. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'Copy Number Ratio CNMF subtypes' versus 'PATHOLOGY.M.STAGE'

P value = 0.864 (Fisher's exact test), Q value = 1

Table S7. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	108	48
subtype1	33	17
subtype2	15	7
subtype3	46	20
subtype4	14	4

Figure S6. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'Copy Number Ratio CNMF subtypes' versus 'GENDER'

P value = 0.0367 (Fisher's exact test), Q value = 1

Table S8. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	121	296
subtype1	35	114
subtype2	19	64
subtype3	54	97
subtype4	13	21

Figure S7. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #7: 'GENDER'

'Copy Number Ratio CNMF subtypes' versus 'HISTOLOGICAL.TYPE'

P value = 0.0989 (Fisher's exact test), Q value = 1

Table S9. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	410	1	6
subtype1	149	0	0
subtype2	80	1	2
subtype3	147	0	4
subtype4	34	0	0

Figure S8. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

'Copy Number Ratio CNMF subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.198 (Fisher's exact test), Q value = 1

Table S10. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	79	338
subtype1	35	114
subtype2	13	70
subtype3	28	123
subtype4	3	31

Figure S9. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'Copy Number Ratio CNMF subtypes' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.000258 (Kruskal-Wallis (anova)), Q value = 0.031

Table S11. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	233	46.8 (37.7)
subtype1	85	49.9 (39.7)
subtype2	51	57.8 (35.0)
subtype3	77	39.6 (38.1)
subtype4	20	33.0 (24.4)

Figure S10. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'Copy Number Ratio CNMF subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.0123 (Kruskal-Wallis (anova)), Q value = 1

Table S12. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	321	2.3 (4.7)
subtype1	123	3.3 (5.7)
subtype2	64	2.0 (5.6)
subtype3	102	1.5 (2.2)
subtype4	32	1.8 (3.0)

Figure S11. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'Copy Number Ratio CNMF subtypes' versus 'RACE'

P value = 0.501 (Fisher's exact test), Q value = 1

Table S13. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	33	361
subtype1	1	4	14	123
subtype2	0	1	10	71
subtype3	0	4	8	136
subtype4	0	1	1	31

Figure S12. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #12: 'RACE'

'Copy Number Ratio CNMF subtypes' versus 'ETHNICITY'

P value = 0.187 (Fisher's exact test), Q value = 1

Table S14. Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	18	381
subtype1	10	134
subtype2	3	75
subtype3	3	141
subtype4	2	31

Figure S13. Get High-res Image Clustering Approach #1: 'Copy Number Ratio CNMF subtypes' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #2: 'METHLYATION CNMF'

Table S15. Description of clustering approach #2: 'METHLYATION CNMF'

Cluster Labels	1	2	3	4	5	6	7
Number of samples	68	58	81	78	58	63	17

'METHLYATION CNMF' versus 'Time to Death'

P value = 0.0801 (logrank test), Q value = 1

Table S16. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	361	101	2.0 - 3837.0 (630.0)
subtype1	60	19	4.0 - 1984.0 (679.5)
subtype2	50	11	37.0 - 2886.0 (628.5)
subtype3	66	19	3.0 - 3837.0 (581.0)
subtype4	65	23	6.0 - 3295.0 (536.0)
subtype5	55	8	10.0 - 2878.0 (721.0)
subtype6	49	13	2.0 - 3835.0 (851.0)
subtype7	16	8	56.0 - 1970.0 (528.5)

Figure S14. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #1: 'Time to Death'

'METHLYATION CNMF' versus 'AGE'

P value = 3.2e-07 (Kruskal-Wallis (anova)), Q value = 4e-05

Table S17. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	422	60.9 (12.2)
subtype1	67	60.1 (13.0)
subtype2	58	57.6 (10.0)
subtype3	81	57.3 (13.5)
subtype4	78	64.2 (11.0)
subtype5	58	57.6 (10.2)
subtype6	63	66.7 (11.0)
subtype7	17	67.8 (11.5)

Figure S15. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #2: 'AGE'

'METHLYATION CNMF' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.83 (Fisher's exact test), Q value = 1

Table S18. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	24	64	71	197	9
subtype1	8	11	12	33	0
subtype2	2	10	11	29	1
subtype3	6	13	16	35	2
subtype4	2	8	11	43	2
subtype5	2	5	10	17	1
subtype6	3	14	8	31	3
subtype7	1	3	3	9	0

Figure S16. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'METHLYATION CNMF' versus 'PATHOLOGY.T.STAGE'

P value = 0.158 (Fisher's exact test), Q value = 1

Table S19. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	40	111	86	133
subtype1	13	21	11	20
subtype2	4	14	14	21
subtype3	8	24	22	19
subtype4	4	17	15	31
subtype5	7	13	9	8
subtype6	3	17	12	27
subtype7	1	5	3	7

Figure S17. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'METHLYATION CNMF' versus 'PATHOLOGY.N.STAGE'

P value = 0.198 (Fisher's exact test), Q value = 1

Table S20. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	137	56	133	7
subtype1	25	11	25	0
subtype2	26	7	16	1
subtype3	26	8	29	1
subtype4	17	9	32	2
subtype5	12	8	12	1
subtype6	27	9	12	2
subtype7	4	4	7	0

Figure S18. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'METHLYATION CNMF' versus 'PATHOLOGY.M.STAGE'

P value = 0.0468 (Fisher's exact test), Q value = 1

Table S21. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	109	48
subtype1	30	5
subtype2	11	9
subtype3	23	9
subtype4	12	7
subtype5	12	9
subtype6	14	9
subtype7	7	0

Figure S19. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'METHLYATION CNMF' versus 'GENDER'

P value = 2e-05 (Fisher's exact test), Q value = 0.0025

Table S22. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	123	300
subtype1	24	44
subtype2	10	48
subtype3	29	52
subtype4	18	60
subtype5	7	51
subtype6	32	31
subtype7	3	14

Figure S20. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #7: 'GENDER'

'METHLYATION CNMF' versus 'HISTOLOGICAL.TYPE'

P value = 0.0464 (Fisher's exact test), Q value = 1

Table S23. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	416	1	6
subtype1	68	0	0
subtype2	57	0	1
subtype3	80	1	0
subtype4	77	0	1
subtype5	54	0	4
subtype6	63	0	0
subtype7	17	0	0

Figure S21. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

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

P value = 0.509 (Fisher's exact test), Q value = 1

Table S24. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	81	342
subtype1	15	53
subtype2	10	48
subtype3	14	67
subtype4	19	59
subtype5	13	45
subtype6	7	56
subtype7	3	14

Figure S22. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'METHLYATION CNMF' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.00134 (Kruskal-Wallis (anova)), Q value = 0.16

Table S25. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	238	46.6 (37.5)
subtype1	42	48.2 (48.3)
subtype2	40	52.1 (28.4)
subtype3	35	34.9 (19.7)
subtype4	55	54.6 (30.9)
subtype5	28	38.1 (40.3)
subtype6	23	51.6 (59.6)
subtype7	15	33.6 (19.3)

Figure S23. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'METHLYATION CNMF' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.000253 (Kruskal-Wallis (anova)), Q value = 0.031

Table S26. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	326	2.3 (4.6)
subtype1	54	2.5 (3.5)
subtype2	50	1.1 (2.4)
subtype3	67	1.9 (2.7)
subtype4	61	4.8 (8.7)
subtype5	26	1.8 (2.3)
subtype6	51	1.2 (2.1)
subtype7	17	2.0 (2.5)

Figure S24. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'METHLYATION CNMF' versus 'RACE'

P value = 0.13 (Fisher's exact test), Q value = 1

Table S27. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	33	367
subtype1	0	4	2	59
subtype2	0	0	8	48
subtype3	1	1	4	71
subtype4	0	2	10	65
subtype5	0	0	3	55
subtype6	0	3	4	55
subtype7	0	0	2	14

Figure S25. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #12: 'RACE'

'METHLYATION CNMF' versus 'ETHNICITY'

P value = 0.698 (Fisher's exact test), Q value = 1

Table S28. Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	18	387
subtype1	1	64
subtype2	2	51
subtype3	6	73
subtype4	4	71
subtype5	2	52
subtype6	2	60
subtype7	1	16

Figure S26. Get High-res Image Clustering Approach #2: 'METHLYATION CNMF' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #3: 'RPPA CNMF subtypes'

Table S29. Description of clustering approach #3: 'RPPA CNMF subtypes'

Cluster Labels	1	2	3	4
Number of samples	55	52	37	68

'RPPA CNMF subtypes' versus 'Time to Death'

P value = 0.647 (logrank test), Q value = 1

Table S30. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	169	75	45.0 - 3837.0 (1000.0)
subtype1	42	21	45.0 - 3837.0 (723.5)
subtype2	45	16	63.0 - 3295.0 (1095.0)
subtype3	30	14	113.0 - 2327.0 (1752.5)
subtype4	52	24	46.0 - 2641.0 (925.0)

Figure S27. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RPPA CNMF subtypes' versus 'AGE'

P value = 0.255 (Kruskal-Wallis (anova)), Q value = 1

Table S31. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	212	62.1 (12.2)
subtype1	55	60.4 (12.3)
subtype2	52	62.7 (9.1)
subtype3	37	62.6 (14.6)
subtype4	68	62.8 (12.9)

Figure S28. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #2: 'AGE'

'RPPA CNMF subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.521 (Fisher's exact test), Q value = 1

Table S32. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	9	39	31	117	4
subtype1	1	14	9	28	1
subtype2	2	4	7	33	1
subtype3	3	8	6	16	1
subtype4	3	13	9	40	1

Figure S29. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'RPPA CNMF subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.17 (Fisher's exact test), Q value = 1

Table S33. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	13	59	53	79
subtype1	3	19	12	19
subtype2	3	6	14	25
subtype3	3	13	6	13
subtype4	4	21	21	22

Figure S30. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'RPPA CNMF subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 0.357 (Fisher's exact test), Q value = 1

Table S34. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	72	21	79	4
subtype1	20	5	19	1
subtype2	15	3	25	1
subtype3	17	5	8	1
subtype4	20	8	27	1

Figure S31. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

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

P value = 0.00683 (Fisher's exact test), Q value = 0.39

Table S35. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #7: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	61	151
subtype1	21	34
subtype2	20	32
subtype3	4	33
subtype4	16	52

Figure S32. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #7: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'RPPA CNMF subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.208 (Kruskal-Wallis (anova)), Q value = 0.74

Table S36. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #9: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	170	2.9 (5.1)
subtype1	42	2.9 (4.2)
subtype2	44	4.5 (8.3)
subtype3	30	1.2 (1.7)
subtype4	54	2.4 (2.8)

Figure S33. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #9: 'NUMBER.OF.LYMPH.NODES'

'RPPA CNMF subtypes' versus 'RACE'

P value = 0.112 (Fisher's exact test), Q value = 1

Table S37. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #10: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	3	19	182
subtype1	1	0	8	44
subtype2	0	1	7	43
subtype3	0	0	1	35
subtype4	0	2	3	60

Figure S34. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #10: 'RACE'

'RPPA CNMF subtypes' versus 'ETHNICITY'

P value = 0.74 (Fisher's exact test), Q value = 1

Table S38. Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #11: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	13	190
subtype1	5	47
subtype2	2	46
subtype3	2	35
subtype4	4	62

Figure S35. Get High-res Image Clustering Approach #3: 'RPPA CNMF subtypes' versus Clinical Feature #11: 'ETHNICITY'

Clustering Approach #4: 'RPPA cHierClus subtypes'

Table S39. Description of clustering approach #4: 'RPPA cHierClus subtypes'

Cluster Labels	1	2	3
Number of samples	78	85	49

'RPPA cHierClus subtypes' versus 'Time to Death'

P value = 0.146 (logrank test), Q value = 1

Table S40. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	169	75	45.0 - 3837.0 (1000.0)
subtype1	62	27	45.0 - 3837.0 (1204.0)
subtype2	67	35	46.0 - 3295.0 (993.0)
subtype3	40	13	77.0 - 2641.0 (980.5)

Figure S36. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RPPA cHierClus subtypes' versus 'AGE'

P value = 0.0759 (Kruskal-Wallis (anova)), Q value = 1

Table S41. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	212	62.1 (12.2)
subtype1	78	61.2 (13.2)
subtype2	85	64.3 (11.8)
subtype3	49	59.9 (10.8)

Figure S37. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'RPPA cHierClus subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.0299 (Fisher's exact test), Q value = 1

Table S42. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	9	39	31	117	4
subtype1	3	18	12	41	1
subtype2	2	11	7	57	2
subtype3	4	10	12	19	1

Figure S38. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'RPPA cHierClus subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.194 (Fisher's exact test), Q value = 1

Table S43. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	13	59	53	79
subtype1	5	26	12	33
subtype2	4	19	27	31
subtype3	4	14	14	15

Figure S39. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'RPPA cHierClus subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 3e-05 (Fisher's exact test), Q value = 0.0037

Table S44. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	72	21	79	4
subtype1	29	9	26	1
subtype2	15	5	45	2
subtype3	28	7	8	1

Figure S40. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

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

P value = 0.436 (Fisher's exact test), Q value = 1

Table S45. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #7: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	61	151
subtype1	22	56
subtype2	28	57
subtype3	11	38

Figure S41. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #7: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'RPPA cHierClus subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 4.76e-06 (Kruskal-Wallis (anova)), Q value = 1

Table S46. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #9: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	170	2.9 (5.1)
subtype1	62	2.5 (3.8)
subtype2	70	4.4 (6.8)
subtype3	38	0.7 (1.3)

Figure S42. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #9: 'NUMBER.OF.LYMPH.NODES'

'RPPA cHierClus subtypes' versus 'RACE'

P value = 0.418 (Fisher's exact test), Q value = 1

Table S47. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #10: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	3	19	182
subtype1	1	0	10	64
subtype2	0	2	6	74
subtype3	0	1	3	44

Figure S43. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #10: 'RACE'

'RPPA cHierClus subtypes' versus 'ETHNICITY'

P value = 0.54 (Fisher's exact test), Q value = 6e-04

Table S48. Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #11: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	13	190
subtype1	7	69
subtype2	4	75
subtype3	2	46

Figure S44. Get High-res Image Clustering Approach #4: 'RPPA cHierClus subtypes' versus Clinical Feature #11: 'ETHNICITY'

Clustering Approach #5: 'RNAseq CNMF subtypes'

Table S49. Description of clustering approach #5: 'RNAseq CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	159	122	127

'RNAseq CNMF subtypes' versus 'Time to Death'

P value = 0.259 (logrank test), Q value = 1

Table S50. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	348	99	2.0 - 3837.0 (643.0)
subtype1	139	36	10.0 - 3835.0 (630.0)
subtype2	104	36	5.0 - 3295.0 (648.0)
subtype3	105	27	2.0 - 3837.0 (725.0)

Figure S45. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq CNMF subtypes' versus 'AGE'

P value = 0.564 (Kruskal-Wallis (anova)), Q value = 1

Table S51. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	407	60.8 (12.2)
subtype1	159	61.1 (10.9)
subtype2	121	59.8 (12.9)
subtype3	127	61.3 (13.1)

Figure S46. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #2: 'AGE'

'RNAseq CNMF subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.141 (Fisher's exact test), Q value = 1

Table S52. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	24	61	68	190	9
subtype1	3	18	28	69	4
subtype2	14	20	17	61	2
subtype3	7	23	23	60	3

Figure S47. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'RNAseq CNMF subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.226 (Fisher's exact test), Q value = 1

Table S53. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	39	105	84	129
subtype1	11	33	32	49
subtype2	18	39	26	32
subtype3	10	33	26	48

Figure S48. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'RNAseq CNMF subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 0.238 (Fisher's exact test), Q value = 1

Table S54. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	132	53	129	7
subtype1	43	19	44	4
subtype2	39	15	52	2
subtype3	50	19	33	1

Figure S49. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'RNAseq CNMF subtypes' versus 'PATHOLOGY.M.STAGE'

P value = 0.447 (Fisher's exact test), Q value = 1

Table S55. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	104	43
subtype1	33	18
subtype2	36	11
subtype3	35	14

Figure S50. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'RNAseq CNMF subtypes' versus 'GENDER'

P value = 0.0124 (Fisher's exact test), Q value = 1

Table S56. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	116	292
subtype1	35	124
subtype2	33	89
subtype3	48	79

Figure S51. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #7: 'GENDER'

'RNAseq CNMF subtypes' versus 'HISTOLOGICAL.TYPE'

P value = 0.00257 (Fisher's exact test), Q value = 0.29

Table S57. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	401	1	6
subtype1	153	0	6
subtype2	121	1	0
subtype3	127	0	0

Figure S52. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

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

P value = 0.286 (Fisher's exact test), Q value = 1

Table S58. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	81	327
subtype1	32	127
subtype2	29	93
subtype3	20	107

Figure S53. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'RNAseq CNMF subtypes' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.015 (Kruskal-Wallis (anova)), Q value = 1

Table S59. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	231	46.6 (37.9)
subtype1	104	53.7 (41.6)
subtype2	61	44.4 (43.5)
subtype3	66	37.6 (20.9)

Figure S54. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'RNAseq CNMF subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.0385 (Kruskal-Wallis (anova)), Q value = 1

Table S60. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	315	2.3 (4.7)
subtype1	99	2.6 (5.7)
subtype2	110	2.8 (5.1)
subtype3	106	1.6 (2.9)

Figure S55. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'RNAseq CNMF subtypes' versus 'RACE'

P value = 0.485 (Fisher's exact test), Q value = 1

Table S61. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	33	352
subtype1	0	2	16	137
subtype2	1	4	7	104
subtype3	0	4	10	111

Figure S56. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #12: 'RACE'

'RNAseq CNMF subtypes' versus 'ETHNICITY'

P value = 0.265 (Fisher's exact test), Q value = 1

Table S62. Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	17	373
subtype1	6	142
subtype2	8	110
subtype3	3	121

Figure S57. Get High-res Image Clustering Approach #5: 'RNAseq CNMF subtypes' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #6: 'RNAseq cHierClus subtypes'

Table S63. Description of clustering approach #6: 'RNAseq cHierClus subtypes'

Cluster Labels	1	2	3	4
Number of samples	42	141	123	102

'RNAseq cHierClus subtypes' versus 'Time to Death'

P value = 0.301 (logrank test), Q value = 1

Table S64. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	348	99	2.0 - 3837.0 (643.0)
subtype1	39	8	156.0 - 2886.0 (644.0)
subtype2	121	42	4.0 - 3295.0 (759.0)
subtype3	98	24	2.0 - 3837.0 (565.5)
subtype4	90	25	10.0 - 2878.0 (681.0)

Figure S58. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'RNAseq cHierClus subtypes' versus 'AGE'

P value = 0.00917 (Kruskal-Wallis (anova)), Q value = 0.96

Table S65. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	407	60.8 (12.2)
subtype1	42	57.1 (8.1)
subtype2	140	59.2 (13.2)
subtype3	123	62.9 (12.5)
subtype4	102	62.0 (11.1)

Figure S59. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'RNAseq cHierClus subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.474 (Fisher's exact test), Q value = 1

Table S66. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	24	61	68	190	9
subtype1	0	6	10	20	1
subtype2	15	21	23	72	2
subtype3	6	23	20	57	3
subtype4	3	11	15	41	3

Figure S60. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.033 (Fisher's exact test), Q value = 1

Table S67. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	39	105	84	129
subtype1	1	9	11	16
subtype2	22	42	34	36
subtype3	9	30	19	52
subtype4	7	24	20	25

Figure S61. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 0.00343 (Fisher's exact test), Q value = 0.39

Table S68. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	132	53	129	7
subtype1	18	5	11	1
subtype2	43	19	61	2
subtype3	54	16	26	1
subtype4	17	13	31	3

Figure S62. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'RNAseq cHierClus subtypes' versus 'PATHOLOGY.M.STAGE'

P value = 0.215 (Fisher's exact test), Q value = 1

Table S69. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	104	43
subtype1	7	6
subtype2	43	11
subtype3	34	17
subtype4	20	9

Figure S63. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'RNAseq cHierClus subtypes' versus 'GENDER'

P value = 0.00103 (Fisher's exact test), Q value = 0.12

Table S70. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	116	292
subtype1	6	36
subtype2	43	98
subtype3	48	75
subtype4	19	83

Figure S64. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #7: 'GENDER'

'RNAseq cHierClus subtypes' versus 'HISTOLOGICAL.TYPE'

P value = 0.0472 (Fisher's exact test), Q value = 1

Table S71. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	401	1	6
subtype1	41	0	1
subtype2	140	0	1
subtype3	123	0	0
subtype4	97	1	4

Figure S65. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

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

P value = 0.0266 (Fisher's exact test), Q value = 1

Table S72. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	81	327
subtype1	6	36
subtype2	34	107
subtype3	15	108
subtype4	26	76

Figure S66. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'RNAseq cHierClus subtypes' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.0442 (Kruskal-Wallis (anova)), Q value = 1

Table S73. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	231	46.6 (37.9)
subtype1	32	53.5 (30.6)
subtype2	69	42.6 (41.6)
subtype3	65	43.0 (38.9)
subtype4	65	51.1 (36.0)

Figure S67. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'RNAseq cHierClus subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 3.29e-05 (Kruskal-Wallis (anova)), Q value = 0.004

Table S74. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	315	2.3 (4.7)
subtype1	35	0.9 (1.7)
subtype2	124	2.6 (4.7)
subtype3	98	1.7 (3.3)
subtype4	58	3.6 (7.0)

Figure S68. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'RNAseq cHierClus subtypes' versus 'RACE'

P value = 0.0491 (Fisher's exact test), Q value = 1

Table S75. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	33	352
subtype1	0	0	7	34
subtype2	1	8	8	118
subtype3	0	1	8	110
subtype4	0	1	10	90

Figure S69. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #12: 'RACE'

'RNAseq cHierClus subtypes' versus 'ETHNICITY'

P value = 0.984 (Fisher's exact test), Q value = 1

Table S76. Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	17	373
subtype1	1	38
subtype2	7	130
subtype3	5	112
subtype4	4	93

Figure S70. Get High-res Image Clustering Approach #6: 'RNAseq cHierClus subtypes' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #7: 'MIRSEQ CNMF'

Table S77. Description of clustering approach #7: 'MIRSEQ CNMF'

Cluster Labels	1	2	3
Number of samples	120	169	132

'MIRSEQ CNMF' versus 'Time to Death'

P value = 0.885 (logrank test), Q value = 1

Table S78. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	359	100	2.0 - 3837.0 (630.0)
subtype1	106	17	2.0 - 2886.0 (442.0)
subtype2	142	41	10.0 - 3837.0 (721.0)
subtype3	111	42	14.0 - 3835.0 (913.0)

Figure S71. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CNMF' versus 'AGE'

P value = 0.269 (Kruskal-Wallis (anova)), Q value = 1

Table S79. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	420	60.9 (12.2)
subtype1	120	62.0 (11.7)
subtype2	169	61.6 (11.2)
subtype3	131	59.1 (13.6)

Figure S72. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #2: 'AGE'

'MIRSEQ CNMF' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.312 (Fisher's exact test), Q value = 1

Table S80. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	24	63	70	197	9
subtype1	8	19	26	49	2
subtype2	5	21	24	83	5
subtype3	11	23	20	65	2

Figure S73. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'MIRSEQ CNMF' versus 'PATHOLOGY.T.STAGE'

P value = 0.141 (Fisher's exact test), Q value = 1

Table S81. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	40	110	85	133
subtype1	12	33	26	35
subtype2	10	35	34	61
subtype3	18	42	25	37

Figure S74. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'MIRSEQ CNMF' versus 'PATHOLOGY.N.STAGE'

P value = 0.193 (Fisher's exact test), Q value = 1

Table S82. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	135	56	133	7
subtype1	44	20	31	2
subtype2	54	16	51	4
subtype3	37	20	51	1

Figure S75. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'MIRSEQ CNMF' versus 'PATHOLOGY.M.STAGE'

P value = 0.16 (Fisher's exact test), Q value = 1

Table S83. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	108	48
subtype1	43	21
subtype2	32	19
subtype3	33	8

Figure S76. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'MIRSEQ CNMF' versus 'GENDER'

P value = 0.372 (Fisher's exact test), Q value = 1

Table S84. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	123	298
subtype1	39	81
subtype2	43	126
subtype3	41	91

Figure S77. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #7: 'GENDER'

'MIRSEQ CNMF' versus 'HISTOLOGICAL.TYPE'

P value = 0.135 (Fisher's exact test), Q value = 1

Table S85. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	414	1	6
subtype1	120	0	0
subtype2	163	1	5
subtype3	131	0	1

Figure S78. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

'MIRSEQ CNMF' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.0555 (Fisher's exact test), Q value = 1

Table S86. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	80	341
subtype1	15	105
subtype2	33	136
subtype3	32	100

Figure S79. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'MIRSEQ CNMF' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.162 (Kruskal-Wallis (anova)), Q value = 1

Table S87. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	237	46.7 (37.6)
subtype1	75	43.7 (37.4)
subtype2	100	50.0 (33.5)
subtype3	62	44.9 (43.8)

Figure S80. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'MIRSEQ CNMF' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.0152 (Kruskal-Wallis (anova)), Q value = 1

Table S88. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	324	2.3 (4.6)
subtype1	92	1.4 (2.4)
subtype2	120	2.5 (5.6)
subtype3	112	2.8 (4.9)

Figure S81. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'MIRSEQ CNMF' versus 'RACE'

P value = 0.291 (Fisher's exact test), Q value = 1

Table S89. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	32	366
subtype1	0	4	9	103
subtype2	0	2	17	148
subtype3	1	4	6	115

Figure S82. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #12: 'RACE'

'MIRSEQ CNMF' versus 'ETHNICITY'

P value = 0.443 (Fisher's exact test), Q value = 1

Table S90. Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	18	385
subtype1	5	110
subtype2	5	154
subtype3	8	121

Figure S83. Get High-res Image Clustering Approach #7: 'MIRSEQ CNMF' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #8: 'MIRSEQ CHIERARCHICAL'

Table S91. Description of clustering approach #8: 'MIRSEQ CHIERARCHICAL'

Cluster Labels	1	2	3
Number of samples	175	99	147

'MIRSEQ CHIERARCHICAL' versus 'Time to Death'

P value = 0.00776 (logrank test), Q value = 0.82

Table S92. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	359	100	2.0 - 3837.0 (630.0)
subtype1	148	49	4.0 - 3295.0 (635.5)
subtype2	83	25	30.0 - 2886.0 (546.0)
subtype3	128	26	2.0 - 3837.0 (709.0)

Figure S84. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #1: 'Time to Death'

'MIRSEQ CHIERARCHICAL' versus 'AGE'

P value = 0.0419 (Kruskal-Wallis (anova)), Q value = 1

Table S93. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	420	60.9 (12.2)
subtype1	174	59.8 (13.2)
subtype2	99	63.5 (11.6)
subtype3	147	60.5 (11.0)

Figure S85. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #2: 'AGE'

'MIRSEQ CHIERARCHICAL' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.121 (Fisher's exact test), Q value = 1

Table S94. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	24	63	70	197	9
subtype1	17	28	34	82	2
subtype2	2	13	14	49	5
subtype3	5	22	22	66	2

Figure S86. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'MIRSEQ CHIERARCHICAL' versus 'PATHOLOGY.T.STAGE'

P value = 0.00483 (Fisher's exact test), Q value = 0.54

Table S95. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	40	110	85	133
subtype1	25	53	42	44
subtype2	3	20	19	42
subtype3	12	37	24	47

Figure S87. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'MIRSEQ CHIERARCHICAL' versus 'PATHOLOGY.N.STAGE'

P value = 0.0716 (Fisher's exact test), Q value = 1

Table S96. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	135	56	133	7
subtype1	56	28	65	2
subtype2	39	7	26	4
subtype3	40	21	42	1

Figure S88. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'MIRSEQ CHIERARCHICAL' versus 'PATHOLOGY.M.STAGE'

P value = 0.204 (Fisher's exact test), Q value = 1

Table S97. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	108	48
subtype1	55	17
subtype2	19	11
subtype3	34	20

Figure S89. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'MIRSEQ CHIERARCHICAL' versus 'GENDER'

P value = 0.0239 (Fisher's exact test), Q value = 1

Table S98. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	123	298
subtype1	54	121
subtype2	37	62
subtype3	32	115

Figure S90. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #7: 'GENDER'

'MIRSEQ CHIERARCHICAL' versus 'HISTOLOGICAL.TYPE'

P value = 0.371 (Fisher's exact test), Q value = 1

Table S99. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	414	1	6
subtype1	173	1	1
subtype2	98	0	1
subtype3	143	0	4

Figure S91. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

'MIRSEQ CHIERARCHICAL' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.127 (Fisher's exact test), Q value = 1

Table S100. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	80	341
subtype1	38	137
subtype2	12	87
subtype3	30	117

Figure S92. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'MIRSEQ CHIERARCHICAL' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.272 (Kruskal-Wallis (anova)), Q value = 1

Table S101. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	237	46.7 (37.6)
subtype1	86	43.9 (38.3)
subtype2	56	51.0 (31.4)
subtype3	95	46.7 (40.3)

Figure S93. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'MIRSEQ CHIERARCHICAL' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.005 (Kruskal-Wallis (anova)), Q value = 0.55

Table S102. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	324	2.3 (4.6)
subtype1	153	2.5 (4.4)
subtype2	79	1.7 (4.1)
subtype3	92	2.6 (5.4)

Figure S94. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'MIRSEQ CHIERARCHICAL' versus 'RACE'

P value = 0.019 (Fisher's exact test), Q value = 1

Table S103. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	32	366
subtype1	1	8	8	150
subtype2	0	2	11	85
subtype3	0	0	13	131

Figure S95. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #12: 'RACE'

'MIRSEQ CHIERARCHICAL' versus 'ETHNICITY'

P value = 0.304 (Fisher's exact test), Q value = 1

Table S104. Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	18	385
subtype1	11	159
subtype2	3	93
subtype3	4	133

Figure S96. Get High-res Image Clustering Approach #8: 'MIRSEQ CHIERARCHICAL' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #9: 'MIRseq Mature CNMF subtypes'

Table S105. Description of clustering approach #9: 'MIRseq Mature CNMF subtypes'

Cluster Labels	1	2	3
Number of samples	128	137	119

'MIRseq Mature CNMF subtypes' versus 'Time to Death'

P value = 0.337 (logrank test), Q value = 1

Table S106. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	324	87	2.0 - 3837.0 (641.5)
subtype1	107	13	3.0 - 2886.0 (436.0)
subtype2	122	35	2.0 - 2562.0 (902.5)
subtype3	95	39	25.0 - 3837.0 (946.0)

Figure S97. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq Mature CNMF subtypes' versus 'AGE'

P value = 0.634 (Kruskal-Wallis (anova)), Q value = 1

Table S107. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	383	61.0 (12.0)
subtype1	127	60.7 (12.2)
subtype2	137	60.5 (11.8)
subtype3	119	61.9 (12.1)

Figure S98. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq Mature CNMF subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.34 (Fisher's exact test), Q value = 1

Table S108. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	23	61	67	173	7
subtype1	7	19	32	50	3
subtype2	8	18	16	62	2
subtype3	8	24	19	61	2

Figure S99. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'MIRseq Mature CNMF subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.217 (Fisher's exact test), Q value = 1

Table S109. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	38	100	80	116
subtype1	14	37	30	32
subtype2	12	23	27	45
subtype3	12	40	23	39

Figure S100. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'MIRseq Mature CNMF subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 0.466 (Fisher's exact test), Q value = 1

Table S110. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	122	54	119	5
subtype1	42	21	37	2
subtype2	46	13	37	2
subtype3	34	20	45	1

Figure S101. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'MIRseq Mature CNMF subtypes' versus 'PATHOLOGY.M.STAGE'

P value = 0.0166 (Fisher's exact test), Q value = 1

Table S111. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	95	48
subtype1	48	35
subtype2	25	4
subtype3	22	9

Figure S102. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'MIRseq Mature CNMF subtypes' versus 'GENDER'

P value = 0.0537 (Fisher's exact test), Q value = 1

Table S112. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	114	270
subtype1	46	82
subtype2	31	106
subtype3	37	82

Figure S103. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #7: 'GENDER'

'MIRseq Mature CNMF subtypes' versus 'HISTOLOGICAL.TYPE'

P value = 0.688 (Fisher's exact test), Q value = 1

Table S113. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	377	1	6
subtype1	124	1	3
subtype2	135	0	2
subtype3	118	0	1

Figure S104. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

'MIRseq Mature CNMF subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.00128 (Fisher's exact test), Q value = 0.15

Table S114. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	62	322
subtype1	9	119
subtype2	30	107
subtype3	23	96

Figure S105. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'MIRseq Mature CNMF subtypes' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.0069 (Kruskal-Wallis (anova)), Q value = 0.74

Table S115. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	210	46.9 (38.7)
subtype1	67	35.6 (19.2)
subtype2	87	57.0 (50.0)
subtype3	56	44.8 (32.0)

Figure S106. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'MIRseq Mature CNMF subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.0614 (Kruskal-Wallis (anova)), Q value = 1

Table S116. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	296	2.3 (4.8)
subtype1	104	1.6 (2.4)
subtype2	94	2.0 (3.9)
subtype3	98	3.4 (6.8)

Figure S107. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'MIRseq Mature CNMF subtypes' versus 'RACE'

P value = 0.0536 (Fisher's exact test), Q value = 1

Table S117. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	30	331
subtype1	1	7	7	108
subtype2	0	2	16	117
subtype3	0	1	7	106

Figure S108. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #12: 'RACE'

'MIRseq Mature CNMF subtypes' versus 'ETHNICITY'

P value = 0.721 (Fisher's exact test), Q value = 1

Table S118. Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	18	349
subtype1	6	115
subtype2	5	126
subtype3	7	108

Figure S109. Get High-res Image Clustering Approach #9: 'MIRseq Mature CNMF subtypes' versus Clinical Feature #13: 'ETHNICITY'

Clustering Approach #10: 'MIRseq Mature cHierClus subtypes'

Table S119. Description of clustering approach #10: 'MIRseq Mature cHierClus subtypes'

Cluster Labels	1	2	3	4	5
Number of samples	73	109	58	95	49

'MIRseq Mature cHierClus subtypes' versus 'Time to Death'

P value = 0.178 (logrank test), Q value = 1

Table S120. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

	nPatients	nDeath	Duration Range (Median), Year
ALL	324	87	2.0 - 3837.0 (641.5)
subtype1	52	12	11.0 - 3837.0 (686.5)
subtype2	94	31	6.0 - 2886.0 (591.5)
subtype3	55	6	2.0 - 2641.0 (717.0)
subtype4	79	30	25.0 - 3295.0 (1030.0)
subtype5	44	8	4.0 - 1969.0 (425.0)

Figure S110. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #1: 'Time to Death'

'MIRseq Mature cHierClus subtypes' versus 'AGE'

P value = 0.807 (Kruskal-Wallis (anova)), Q value = 1

Table S121. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'AGE'

	nPatients	Mean (Std.Dev)
ALL	383	61.0 (12.0)
subtype1	73	62.0 (13.8)
subtype2	109	61.6 (10.1)
subtype3	58	60.2 (10.7)
subtype4	95	60.8 (12.9)
subtype5	48	59.7 (13.2)

Figure S111. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #2: 'AGE'

'MIRseq Mature cHierClus subtypes' versus 'NEOPLASM.DISEASESTAGE'

P value = 0.00161 (Fisher's exact test), Q value = 0.19

Table S122. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

nPatients	STAGE I	STAGE II	STAGE III	STAGE IVA	STAGE IVB
ALL	23	61	67	173	7
subtype1	0	15	12	36	3
subtype2	3	13	17	59	2
subtype3	4	7	11	14	1
subtype4	10	13	16	51	0
subtype5	6	13	11	13	1

Figure S112. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #3: 'NEOPLASM.DISEASESTAGE'

'MIRseq Mature cHierClus subtypes' versus 'PATHOLOGY.T.STAGE'

P value = 0.00231 (Fisher's exact test), Q value = 0.27

Table S123. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

nPatients	T1	T2	T3	T4
ALL	38	100	80	116
subtype1	2	21	12	32
subtype2	7	20	27	40
subtype3	8	14	7	10
subtype4	13	27	23	27
subtype5	8	18	11	7

Figure S113. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #4: 'PATHOLOGY.T.STAGE'

'MIRseq Mature cHierClus subtypes' versus 'PATHOLOGY.N.STAGE'

P value = 0.0382 (Fisher's exact test), Q value = 1

Table S124. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

nPatients	N0	N1	N2	N3
ALL	122	54	119	5
subtype1	28	10	18	1
subtype2	32	12	40	2
subtype3	15	10	9	1
subtype4	24	16	41	0
subtype5	23	6	11	1

Figure S114. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #5: 'PATHOLOGY.N.STAGE'

'MIRseq Mature cHierClus subtypes' versus 'PATHOLOGY.M.STAGE'

P value = 0.32 (Fisher's exact test), Q value = 1

Table S125. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

nPatients	M0	MX
ALL	95	48
subtype1	18	13
subtype2	19	11
subtype3	13	9
subtype4	22	5
subtype5	23	10

Figure S115. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #6: 'PATHOLOGY.M.STAGE'

'MIRseq Mature cHierClus subtypes' versus 'GENDER'

P value = 0.424 (Fisher's exact test), Q value = 1

Table S126. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #7: 'GENDER'

nPatients	FEMALE	MALE
ALL	114	270
subtype1	24	49
subtype2	25	84
subtype3	19	39
subtype4	32	63
subtype5	14	35

Figure S116. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #7: 'GENDER'

'MIRseq Mature cHierClus subtypes' versus 'HISTOLOGICAL.TYPE'

P value = 0.179 (Fisher's exact test), Q value = 1

Table S127. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

nPatients	HEAD AND NECK SQUAMOUS CELL CARCINOMA	HEAD AND NECK SQUAMOUS CELL CARCINOMA SPINDLE CELL VARIANT	HEAD AND NECK SQUAMOUS CELL CARCINOMA BASALOID TYPE
ALL	377	1	6
subtype1	72	1	0
subtype2	107	0	2
subtype3	55	0	3
subtype4	94	0	1
subtype5	49	0	0

Figure S117. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #8: 'HISTOLOGICAL.TYPE'

'MIRseq Mature cHierClus subtypes' versus 'RADIATIONS.RADIATION.REGIMENINDICATION'

P value = 0.0869 (Fisher's exact test), Q value = 1

Table S128. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

nPatients	NO	YES
ALL	62	322
subtype1	9	64
subtype2	19	90
subtype3	9	49
subtype4	22	73
subtype5	3	46

Figure S118. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #9: 'RADIATIONS.RADIATION.REGIMENINDICATION'

'MIRseq Mature cHierClus subtypes' versus 'NUMBERPACKYEARSSMOKED'

P value = 0.0392 (Kruskal-Wallis (anova)), Q value = 1

Table S129. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

	nPatients	Mean (Std.Dev)
ALL	210	46.9 (38.7)
subtype1	34	41.6 (23.7)
subtype2	72	54.3 (34.8)
subtype3	33	42.4 (51.4)
subtype4	41	48.0 (51.4)
subtype5	30	38.8 (19.1)

Figure S119. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #10: 'NUMBERPACKYEARSSMOKED'

'MIRseq Mature cHierClus subtypes' versus 'NUMBER.OF.LYMPH.NODES'

P value = 0.0412 (Kruskal-Wallis (anova)), Q value = 1

Table S130. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

	nPatients	Mean (Std.Dev)
ALL	296	2.3 (4.8)
subtype1	60	2.2 (3.8)
subtype2	83	2.7 (6.2)
subtype3	29	1.4 (2.0)
subtype4	83	3.0 (5.4)
subtype5	41	1.0 (1.4)

Figure S120. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #11: 'NUMBER.OF.LYMPH.NODES'

'MIRseq Mature cHierClus subtypes' versus 'RACE'

P value = 0.00655 (Fisher's exact test), Q value = 0.71

Table S131. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #12: 'RACE'

nPatients	AMERICAN INDIAN OR ALASKA NATIVE	ASIAN	BLACK OR AFRICAN AMERICAN	WHITE
ALL	1	10	30	331
subtype1	0	1	5	64
subtype2	0	2	14	91
subtype3	0	1	7	50
subtype4	0	1	3	87
subtype5	1	5	1	39

Figure S121. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #12: 'RACE'

'MIRseq Mature cHierClus subtypes' versus 'ETHNICITY'

P value = 0.248 (Fisher's exact test), Q value = 1

Table S132. Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #13: 'ETHNICITY'

nPatients	HISPANIC OR LATINO	NOT HISPANIC OR LATINO
ALL	18	349
subtype1	5	65
subtype2	4	97
subtype3	0	56
subtype4	6	88
subtype5	3	43

Figure S122. Get High-res Image Clustering Approach #10: 'MIRseq Mature cHierClus subtypes' versus Clinical Feature #13: 'ETHNICITY'

Methods & Data

Input

Cluster data file = HNSC-TP.mergedcluster.txt
Clinical data file = HNSC-TP.merged_data.txt
Number of patients = 423
Number of clustering approaches = 10
Number of selected clinical features = 13
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

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

Q value calculation

For multiple hypothesis correction, Q value is the False Discovery Rate (FDR) analogue of the P value (Benjamini and Hochberg 1995), defined as the minimum FDR at which the test may be called significant. We used the 'Benjamini and Hochberg' method of 'p.adjust' function in R to convert P values into Q values.

Download Results

In addition to the links below, the full results of the analysis summarized in this report can also be downloaded programmatically using firehose_get, or interactively from either the Broad GDAC website or TCGA Data Coordination Center Portal.

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] 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)

[5] Benjamini and Hochberg, Controlling the false discovery rate: a practical and powerful approach to multiple testing, Journal of the Royal Statistical Society Series B 59:289-300 (1995)

Made with Nozzle