Skip to content
LB2

Brain lower grade glioma LGG

Of 60,498 genes, 3,084 are higher in the tumor than in normal brain and in whole blood, and are not made by blood immune cells. 1,181 are protein-coding, and 1,171 of those have supporting plasma Cell-free RNA (cfRNA)RNA fragments that circulate in blood plasma outside cells. Most come from blood cells; a small share comes from other tissues, including tumors. evidence; the rest are mostly non-coding genes the plasma sources cannot check.

Samples compared

Tumor
509
TCGA primary tumors
Normal tissue
1,152
1,152 GTEx brain
Whole blood
337
GTEx samples from 328 donors
Survival
504
patients, 189 events (Survival endpointsCurated TCGA outcomes (Liu et al. 2018): OS is overall survival, PFI the progression-free interval and DSS disease-specific survival. LB2 uses whichever has the most events in each cancer type.)

Results in brief

3,084
candidate genes: 1,171 with plasma cfRNA evidence, 20 protein-coding without, 1,893 non-coding
242
genes in the tissue panel, Held-out AUCHow well an elastic-net model separates tumor from normal tissue, measured with nested cross-validation on samples the model did not train on. It describes tissue, not the accuracy of a blood test. 0.998
62
genes associated with progression-free interval (FDR < 0.05, 150 tested)

Candidates

Filter by plasma evidence or discrimination, find a gene, and select any point or row for its full record. The table holds all 3,084 candidates.

Loading candidates…

How the candidates were selected

60,498 genes tested; 8,362 higher in tumor than in normal brain; 7,169 also higher than in whole blood; 3,084 remain after removing genes expressed in blood immune cells; 1,171 of these have plasma cfRNA evidence, 20 protein-coding candidates have none, and 1,893 are non-coding genes the plasma sources cannot list.

Figure 2. Genes remaining after each step in brain lower grade glioma.

Each comparison uses a two-sided Wilcoxon rank-sum test. A gene passes when its q-value is below 0.05, its log2 fold change is at least 1, and the lower bound of the 95% confidence interval for its AUC is at least 0.70.

Genes made by any of 18 sorted blood immune cell types above 1 nTPM are then removed, because blood cells supply most plasma RNA. Genes missing from that reference, mostly non-coding, are kept.

Plasma evidence only reorders the list. A candidate not yet seen in plasma stays in, because every dataset misses genes. Read the full methods.

Tissue classifier panel

Genetoward normaltoward tumorCoef.
  1. NACA2+0.29
  2. RCOR2+0.28
  3. PBK+0.25
  4. EGFR+0.25
  5. BMP2+0.24
  6. RFX4−0.23
  7. MEX3A+0.23
  8. MMP2+0.23
  9. PCDHB9+0.22
  10. NFIB+0.21
  11. BCHE+0.21
  12. TBX15+0.20
  13. MDFI+0.20
  14. ANGPTL2+0.20
  15. CBX2+0.20
  16. ATOH8+0.20
  17. MMP15+0.20
  18. SIX1+0.20
  19. TSKU+0.19
  20. HAPLN1+0.19
Show the other 222 genes
  1. NES+0.18
  2. C1QL1+0.18
  3. MMP16+0.18
  4. SPSB4+0.18
  5. TSPAN11+0.18
  6. FJX1+0.17
  7. TSPAN12+0.17
  8. CCDC80+0.17
  9. MSI1+0.17
  10. NLGN1+0.17
  11. FERMT1+0.17
  12. CSPG5+0.17
  13. SOX8+0.17
  14. GPC2+0.16
  15. IL17RD+0.16
  16. TIMP4+0.16
  17. DCX+0.16
  18. C1orf106+0.15
  19. LRP5+0.15
  20. GAS1+0.15
  21. TRIM24+0.15
  22. TMEM100+0.14
  23. TMEM108+0.14
  24. LRRN1+0.14
  25. TTC30A+0.14
  26. CYP27C1+0.13
  27. SAPCD2+0.13
  28. PDGFC+0.13
  29. C14orf37+0.13
  30. NFIX+0.13
  31. BMP7+0.13
  32. SPC25+0.13
  33. DDR1+0.13
  34. SRGAP1−0.13
  35. PIK3R2+0.12
  36. TICRR−0.12
  37. KIAA1549+0.12
  38. FAM181B+0.12
  39. LIFR+0.12
  40. TIMP3+0.12
  41. POU3F3−0.12
  42. GLIS2+0.11
  43. FLRT1+0.11
  44. VANGL2+0.11
  45. CHST9+0.11
  46. TNN+0.11
  47. CSRP2+0.11
  48. DUS4L−0.10
  49. ST8SIA1−0.10
  50. MANEAL+0.10
  51. TMEM5−0.10
  52. NOVA1−0.10
  53. MASP1+0.10
  54. ZNF462−0.10
  55. DCLK2+0.10
  56. ZFHX4−0.10
  57. BEST3+0.10
  58. PHF21B−0.09
  59. PDGFRA+0.09
  60. QSER1+0.09
  61. MTTP−0.09
  62. ZFHX3−0.09
  63. NRP2−0.09
  64. APCDD1+0.09
  65. PRTG−0.09
  66. SLC4A4+0.08
  67. HSF2BP+0.08
  68. AGT+0.08
  69. ANTXR1+0.08
  70. ANOS1+0.08
  71. CRISPLD1+0.08
  72. ZNF629+0.08
  73. TMSB15A+0.08
  74. ZBTB10+0.08
  75. ASCL1+0.08
  76. DSEL+0.07
  77. SEMA6A−0.07
  78. LRP4+0.07
  79. FERMT2−0.07
  80. PCDHB10+0.07
  81. ZNF501+0.07
  82. ZSWIM4+0.07
  83. ZBED3+0.07
  84. SPARCL1+0.07
  85. FRMPD3+0.07
  86. GAREML−0.07
  87. EDNRA+0.06
  88. GRIK3+0.06
  89. KIAA1211+0.06
  90. MSX1+0.06
  91. GIPC3+0.06
  92. TJP1−0.06
  93. SDC3+0.06
  94. TOX3+0.06
  95. AMOTL2+0.06
  96. RGMA+0.06
  97. NCAM2−0.06
  98. POU3F2−0.06
  99. ASTN1−0.06
  100. SOWAHC−0.06
  101. C1orf94+0.06
  102. SCARF2+0.05
  103. SLC16A2+0.05
  104. NFKBIL1+0.05
  105. WSCD1+0.05
  106. NCAN+0.05
  107. PGF+0.05
  108. SOX21+0.05
  109. WNT5A−0.05
  110. HTRA1+0.05
  111. NEU4+0.05
  112. PLCD3+0.05
  113. PHYHIPL+0.05
  114. PTPRZ1+0.05
  115. TNR−0.05
  116. SH3RF3+0.04
  117. ID4+0.04
  118. SOX11+0.04
  119. HEY2−0.04
  120. GPR161+0.04
  121. PCDH18+0.04
  122. COX18−0.04
  123. C1orf226+0.04
  124. TTC28−0.04
  125. MYCN+0.04
  126. KCNJ10−0.04
  127. DPYSL3+0.04
  128. MTHFD2L−0.04
  129. MIB1+0.04
  130. IGFBP5+0.04
  131. RND2+0.04
  132. PTPRG+0.04
  133. ARHGAP42+0.04
  134. RFTN2+0.04
  135. VARS+0.04
  136. CTNND2+0.03
  137. MICALL1−0.03
  138. LRRC1−0.03
  139. SLCO5A1+0.03
  140. EDNRB+0.03
  141. MTSS1L−0.03
  142. PHKG1−0.03
  143. JAM2+0.03
  144. TMEM99+0.03
  145. DGCR14+0.03
  146. SALL1−0.03
  147. DSCAM−0.03
  148. ZXDA+0.03
  149. ADCYAP1R1+0.03
  150. TRIM16L−0.03
  151. DOK5+0.03
  152. PCDH17+0.03
  153. HGH1+0.03
  154. GTF2IRD1−0.03
  155. CRMP1−0.03
  156. SLCO2B1−0.03
  157. LLGL1−0.03
  158. DHFR+0.03
  159. CRYAB+0.03
  160. C1QL4+0.03
  161. CDC25A+0.03
  162. GLUD2+0.03
  163. KCNN2+0.03
  164. ADGRL3−0.03
  165. PRRX1−0.02
  166. SMAD9+0.02
  167. C6orf48−0.02
  168. PCDHB16+0.02
  169. SULT1C4−0.02
  170. ADAMTS12+0.02
  171. STC2+0.02
  172. C8orf37−0.02
  173. CENPBD1+0.02
  174. FAM69C+0.02
  175. C19orf48+0.02
  176. NKD1+0.02
  177. GRHL3+0.02
  178. NIM1K+0.02
  179. ZBTB5−0.02
  180. SPIRE1+0.02
  181. ZBTB22+0.02
  182. ZKSCAN2+0.02
  183. FGF11+0.02
  184. PXDN−0.02
  185. URB1−0.02
  186. ZNF853−0.01
  187. C22orf23−0.01
  188. ZDHHC15+0.01
  189. CD276+0.01
  190. GPC4−0.01
  191. DNER+0.01
  192. GRIK4+0.01
  193. GSX1+0.01
  194. CXXC4+0.01
  195. PBX1+0.01
  196. CTD-2192J16.20+0.01
  197. EBF4+0.01
  198. TENM4−0.01
  199. NEDD4−0.01
  200. ZNF707−0.01
  201. HNF4G−0.01
  202. TXNRD3−0.01
  203. DLG5+0.01
  204. UBD+0.01
  205. SMO+0.01
  206. KIF7+0.01
  207. HAS2−0.01
  208. AKAP3+0.01
  209. CSPG4+0.01
  210. SNX33−0.01
  211. C1orf61−0.01
  212. ADGRA3+0.01
  213. NAALADL2+0.00
  214. HOMEZ−0.00
  215. MOCS1−0.00
  216. PARD3B−0.00
  217. ADAMTS15+0.00
  218. LUZP2+0.00
  219. TNC+0.00
  220. NTN1+0.00
  221. PVRL3+0.00
  222. ROBO4+0.00
Figure 3. Standardized elastic-net coefficients of the 242 genes selected to separate brain lower grade glioma from normal brain, largest first.

An elastic-net logistic regression was trained on the 300 top-ranked candidates to tell tumor from normal tissue. Its Held-out AUCHow well an elastic-net model separates tumor from normal tissue, measured with nested cross-validation on samples the model did not train on. It describes tissue, not the accuracy of a blood test. is 0.998, from nested five-fold cross-validation.

Every normal sample here comes from GTEx, so the TCGA-versus-GTEx offset cannot be separated from the tumor/normal contrast and no study correction was possible: the classifier may partly learn study differences.

This number describes tumor and normal tissue, where separation is expected to be near perfect. It is not the accuracy of a blood test: that has to be measured in plasma from patients and controls.

Association with survival

Gene0.250.5124HR per SD (95% CI)q-value
  1. TNC1.86 (1.59–2.19)1.5 × 10−12
  2. CD2761.65 (1.43–1.90)1.8 × 10−10
  3. HNF4G1.55 (1.31–1.83)3.4 × 10−6
  4. SMO1.54 (1.31–1.81)1.9 × 10−6
  5. CRISPLD11.51 (1.30–1.75)6.2 × 10−7
  6. NES1.49 (1.27–1.75)8.5 × 10−6
  7. MMP21.47 (1.26–1.72)6.9 × 10−6
  8. GAS11.47 (1.28–1.67)2.0 × 10−7
  9. CDH111.45 (1.25–1.67)6.9 × 10−6
  10. C1orf610.73 (0.64–0.83)1.1 × 10−5
  11. C1QL10.71 (0.62–0.82)1.7 × 10−5
  12. TMEM1000.69 (0.61–0.79)2.0 × 10−7
  13. NFKBIL10.68 (0.59–0.80)7.4 × 10−6
  14. ATAT10.68 (0.58–0.79)1.0 × 10−5
  15. ANGPTL20.67 (0.59–0.77)1.4 × 10−7
  16. ATOH80.65 (0.57–0.75)4.5 × 10−8
  17. SOX80.65 (0.57–0.73)2.2 × 10−10
  18. HSF2BP0.64 (0.55–0.75)3.1 × 10−7
  19. FERMT10.62 (0.55–0.71)9.9 × 10−12
  20. BMP20.53 (0.46–0.60)9.6 × 10−18
Show the other 42 genes
  1. SALL11.47 (1.21–1.79)6.9 × 10−4
  2. POU3F21.42 (1.18–1.70)8.3 × 10−4
  3. SNX71.39 (1.20–1.62)6.9 × 10−5
  4. MSI11.39 (1.18–1.64)4.7 × 10−4
  5. SRGAP11.39 (1.17–1.65)8.3 × 10−4
  6. PTPRZ11.35 (1.14–1.59)0.0021
  7. EGFR1.34 (1.14–1.58)0.0020
  8. WNT5A1.34 (1.16–1.55)4.2 × 10−4
  9. TRIM241.33 (1.15–1.55)8.5 × 10−4
  10. TSPAN121.32 (1.12–1.55)0.0037
  11. SDC31.31 (1.13–1.52)0.0019
  12. SOWAHC1.31 (1.11–1.53)0.0041
  13. PCDH181.23 (1.06–1.43)0.018
  14. CBX21.23 (1.05–1.44)0.031
  15. SNX331.23 (1.05–1.43)0.029
  16. CLIP21.23 (1.04–1.44)0.036
  17. TTC30A1.22 (1.05–1.42)0.026
  18. NTN11.22 (1.03–1.43)0.043
  19. ARHGAP421.22 (1.04–1.43)0.041
  20. DSEL1.21 (1.04–1.42)0.040
  21. ZDHHC151.21 (1.03–1.42)0.043
  22. ZNF6291.20 (1.03–1.39)0.045
  23. DLG50.85 (0.75–0.96)0.026
  24. FAM181B0.84 (0.74–0.97)0.041
  25. DCLK20.84 (0.73–0.97)0.043
  26. MYCN0.83 (0.71–0.96)0.035
  27. KIAA15490.83 (0.72–0.95)0.026
  28. PRTG0.82 (0.72–0.94)0.017
  29. PCDHB90.82 (0.71–0.96)0.031
  30. RFTN20.81 (0.70–0.94)0.014
  31. NKD10.81 (0.70–0.93)0.013
  32. DGCR140.80 (0.69–0.94)0.019
  33. SPIRE10.79 (0.69–0.91)0.0031
  34. RND20.77 (0.68–0.89)8.5 × 10−4
  35. EBF40.77 (0.67–0.89)0.0011
  36. FGF110.77 (0.67–0.88)8.3 × 10−4
  37. GRIK40.76 (0.66–0.88)8.3 × 10−4
  38. ADGRL30.76 (0.65–0.87)8.2 × 10−4
  39. NFIB0.76 (0.65–0.87)8.3 × 10−4
  40. LRRN10.75 (0.65–0.86)4.7 × 10−4
  41. ADCYAP1R10.72 (0.62–0.84)2.7 × 10−4
  42. NKAIN40.72 (0.63–0.82)1.7 × 10−5
longer PFIshorter PFI
Figure 4. Hazard ratio per standard deviation of tumor expression with 95% confidence interval, log scale, for the 62 genes with FDR < 0.05.

For the 150 top-ranked candidates, a Cox model relates tumor expression, as a continuous value, to progression-free interval in 504 patients (189 events). No high/low cutpoint is searched for, since optimized cutpoints inflate false positives.

A Hazard ratio per SDFrom a Cox model with the gene’s tumor expression as a continuous variable: the change in hazard for each one-standard-deviation increase. Above 1, higher expression goes with a shorter time to the event; below 1, with a longer one. above 1 means higher expression goes with a shorter progression-free interval. These are associations in tissue, not evidence that a blood level predicts outcome.

Caveats for LGG

  • Every normal sample comes from GTEx, a different study from the tumors, so study differences cannot be corrected in the tissue comparison.
  • The blood comparison sets TCGA tumors against GTEx blood. Study and biology cannot be separated there, which is why genes made by blood immune cells are removed as well.

Limitations that apply to every cancer type