Skip to content
LB2

Candidate cell-free RNA markers for 29 cancer types

LB2 looks for genes that are expressed more in a tumor than in its normal tissue and in blood cells: the pattern a tumor-derived transcript needs to stand out in plasma. It compares 9,059 TCGA tumors with normal tissue and 337 GTEx whole-blood samples, then checks each hit against published 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. data.

Try FOXM1, CEACAM5 or pancreatic adenocarcinoma.

These are candidates for testing in patient plasma, not validated biomarkers. What the screen cannot tell you

60,498 genes tested; 3,560 higher in tumor than in normal lung; 1,957 also higher than in whole blood; 1,364 remain after removing genes expressed in blood immune cells; 624 of these have plasma cfRNA evidence, 13 protein-coding candidates have none, and 727 are non-coding genes the plasma sources cannot list.

Figure 1. From 60,498 genes to 1,364 candidates in lung adenocarcinoma. Each comparison keeps genes with q < 0.05, a log2 fold change of at least 1 and a lower 95% AUC bound of at least 0.70. Plasma evidence is recorded, never used to remove genes; the two cfRNA datasets cover protein-coding genes only. Explore LUAD

Browse by cancer type

48,527 candidates in 30 cancer types, from 14,761 distinct genes. Bar length is the number of candidates; the split shows how many have plasma cfRNA evidence (21,254 overall), how many protein-coding candidates have none yet (381), and how many are non-coding genes the plasma sources cannot check (26,892).

plasma evidencecoding, not detectednon-codingCompare in a table

Thoracic

Gastrointestinal

Genitourinary

Gynecologic

Breast

Endocrine

Central nervous system

Head and neck

Skin

Soft tissue

Eye

  • UVMUveal melanomaNot analysed. No matched normal tissue in TCGA or GTEx.–

Lymphoid

  • DLBCLymphoid neoplasm diffuse large B-cell lymphomaNot analysed. No matched normal tissue in TCGA or GTEx.–

Why compare with both normal tissue and blood

Most RNA in plasma comes from blood cells, so a tumor marker has to be higher in the tumor than in blood. On its own, that comparison mostly finds genes that mark the organ rather than the cancer.

NKX2-1, the lung marker pathologists stain for as TTF-1, is far higher in lung adenocarcinoma than in blood, yet no higher than in normal lung. LB2 therefore also requires a gene to be higher than in the tumor’s own normal tissue.

MMP9 is higher in tumors than in normal lung, but blood carries even more of it, made by neutrophils. SPP1 passes both comparisons, yet sorted neutrophils express it too, so the immune-cell filter removes it.

CEACAM5, the gene behind the clinical serum marker CEA, passes every step and is supported by all three plasma sources. How each step works

Five lung adenocarcinoma genes placed by their two comparisonsCEACAM5 and FAM83A are higher than both normal lung and blood and are candidates. NKX2-1 is much higher than blood but not higher than normal lung. MMP9 is higher than normal lung but not than blood. SPP1 passes both comparisons but is expressed in neutrophils, so it is removed.Higher than bothHigher than blood onlymarks the organ,not the cancerHigher than normal tissue onlymasked by blood cells−4−20+2+4+6+8−40+4+8+12+16log2 fold change, tumor vs normal lunglog2 fold change, tumor vs whole bloodCEACAM5FAM83ANKX2-1MMP9SPP1
Figure 2. Five genes from lung adenocarcinoma, placed by their two comparisons. Dashed lines mark a twofold difference; the shaded region is where candidates can fall.