Bile Sequencing Turns Discarded Fluid Into Early Cancer Clues

Every year, thousands of patients undergo endoscopic retrograde cholangiopancreatography, or ERCP, because imaging has revealed a narrowing of the bile duct that doctors cannot immediately explain. These biliary strictures are among the most diagnostically frustrating lesions in gastroenterology. Some are caused by aggressive cancers such as cholangiocarcinoma or pancreatic ductal adenocarcinoma, while others are the benign scars of chronic inflammation, gallstone disease or previous surgery. Telling the two apart quickly matters enormously, because early-stage pancreatobiliary cancers offer the only realistic window for curative surgery, yet the standard diagnostic tools frequently come back inconclusive.

The problem is rooted in the biology and anatomy of these tumours. Brush cytology and forceps biopsy taken during ERCP sample only a small fraction of the stricture, and desmoplastic stroma often surrounds sparse malignant cells, so malignant strictures can yield false negatives. Serum carbohydrate antigen 19-9, the most widely used blood marker, is also unreliable in this setting: cholestasis itself, the bile-flow obstruction that brings patients to endoscopy in the first place, can push CA19-9 concentrations above the usual 44 U/mL cutoff even when no cancer is present. The result is a diagnostic limbo in which patients undergo repeated procedures, delayed referrals and, in the worst cases, missed opportunities for resection.

A study published in eGastroenterology by Rojo and colleagues at the First Hospital of Jilin University proposes a strikingly simple way to extract more information from a procedure patients already undergo: sequence the bile. Bile collected during routine ERCP is ordinarily treated as waste, but it bathes the tumour and is rich in cell-free DNA, fragments of genetic material shed by dying cells. Because malignant cells in the ductal epithelium and pancreas release their DNA directly into this fluid, bile offers a far more concentrated and anatomically relevant source of tumour DNA than peripheral blood, where pancreatobiliary cancer DNA is heavily diluted.

The technical elegance of the approach lies in its economy. Rather than deep sequencing, the team used ultra-low-pass whole-genome sequencing, or ULP-WGS, reading the bile cell-free DNA at roughly 0.2× genomic coverage from as little as 2 nanograms of starting material. At such shallow depth, individual bases cannot be reliably called, but the method does something arguably more useful: it reveals large-scale copy-number alterations, the gains and losses of whole chromosome arms that are hallmarks of cancer genomes. Using the open-source ichorCNA pipeline, the researchers computed a tumour fraction, the estimated proportion of cell-free DNA derived from tumour, with a threshold of 10 per cent or more indicating detectable circulating tumour DNA. Reducing sequencing depth from the conventional 30–50× used in clinical whole-genome sequencing to 0.2× was estimated to cut sequencing costs by approximately 130- to 215-fold, a difference that could make the test scalable in routine practice.

The prospective cohort comprised 95 patients undergoing their first ERCP for a biliary stricture that was radiologically suspicious for malignancy. At the outset, pathological assessment classified 26 strictures as malignant and 69 as indeterminate. Over follow-up of up to 12 months, 45 of the initially indeterminate cases were confirmed malignant and 24 proved benign, giving a final malignant cohort of 40 cholangiocarcinomas and 31 pancreatic ductal adenocarcinomas. This design mirrors the real clinical dilemma, since the patients who matter most are precisely those in whom the first pathological evaluation fails to settle the question.

The performance figures are encouraging on both ends of the spectrum. Among patients already diagnosed with malignancy at the first ERCP, bile ULP-WGS detected tumour DNA in 73 per cent of cases, including 82 per cent of cholangiocarcinomas and 56 per cent of pancreatic cancers. More importantly, in the initially indeterminate strictures later confirmed as malignant, sensitivity was 51 per cent, while 23 of the 24 ultimately benign cases tested negative, corresponding to a specificity of 96 per cent. By comparison, CA19-9 at the prespecified threshold of 44 U/mL achieved only 53 per cent specificity in evaluable benign cases, underscoring how poorly the conventional marker performs in the obstructed, inflamed patients who need clarity most.

The most clinically compelling finding concerns timing. In the 23 patients whose malignancy was not established at the first ERCP but whose bile tested positive, tumour DNA was detected a median of 38 days before the conventional diagnosis was made. In seven patients the genomic signal preceded diagnosis by more than 100 days, and in one case by an extraordinary 499 days. For cancers in which every week of delay can shift a patient from resectable to unresectable disease, a lead time measured in months could translate directly into surgical candidacy and survival. The researchers also compared their shallow approach with a targeted next-generation sequencing assay of 52 genes, dubbed Bilemut, which detected mutations in 94.7 per cent of evaluable malignant strictures and remained more sensitive overall. ULP-WGS, however, required less DNA and delivered genome-wide copy-number information that targeted mutation panels cannot capture, suggesting the two methods are complementary rather than competing.

Beyond detecting cancer, the copy-number profiles carried information about where the tumour had arisen. The bile-derived alterations reproduced genomic patterns previously described in pancreatobiliary tumour tissue, including recurrent losses of chromosome arms 9p, 17p and 18q and gains of 7p, 7q, 8q and 19q. A supervised classifier built on 50 copy-number regions distinguished cholangiocarcinoma from pancreatic ductal adenocarcinoma with 86.4 per cent accuracy in the reference cohort and 85.0 per cent in a separate validation cohort. Loss of 1p was enriched in pancreatic cancer, whereas 9p loss occurred more frequently in cholangiocarcinoma. This matters because the two tumours demand different management, from neoadjuvant strategies to surgical planning, and tissue-of-origin assignment from a liquid biopsy could guide decisions even before a formal histological diagnosis exists.

The sequencing data also appeared to carry prognostic weight. Neither tumour fraction alone nor targeted tumour mutational burden alone significantly stratified overall survival, but the combination did. Patients who were both tumour-fraction positive and carried mutations in at least three genes had a median overall survival of 5.1 months, compared with 13.7 months among the remaining patients. After adjustment for initial management, CA19-9 and carcinoembryonic antigen, this combined genomic profile remained independently associated with poorer survival, with an adjusted hazard ratio of 2.51 and a 95 per cent confidence interval of 1.22 to 5.14. If validated, such a signature could help clinicians identify patients whose disease course is likely to be aggressive and for whom early referral to palliative or intensified protocols might be appropriate.

The authors are careful about the limits of their test. A negative bile ULP-WGS result is not sensitive enough to exclude malignancy, and the method should not replace histopathology, which remains the diagnostic gold standard. Its most promising role is as a complementary test after an inconclusive first ERCP: a positive result could justify accelerated repeat sampling, cholangioscopy with directed biopsy, endoscopic ultrasound-guided tissue acquisition or closer multidisciplinary surveillance, while sparing patients with negative results some of the cascade of repeat procedures. The study also has constraints that temper enthusiasm. It was conducted at a single centre, the cohort was modest, the ultimately benign group was small, and the cholangiocarcinoma-versus-pancreatic-cancer classifier requires external validation before it can inform real-world decisions. Prospective multicentre studies must now determine whether bile genomics can genuinely shorten diagnostic delays, reduce the number of invasive repeat procedures and, ultimately, improve outcomes for patients caught in the diagnostic grey zone of the biliary stricture. What the current work establishes, however, is that a fluid routinely discarded at the end of every ERCP contains a readable, cost-effective genomic record of the tumours it touches, and that reading it may move pancreatobiliary cancer diagnosis weeks or even months earlier.

Subject of Research: Ultra-low-pass whole-genome sequencing of bile cell-free DNA for detecting and classifying malignant biliary strictures

Article Title: Bile genome sequencing may accelerate diagnosis of malignant biliary strictures

Article References: Bile genome sequencing may accelerate diagnosis of malignant biliary strictures. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: bile cell-free DNA, ultra-low-pass whole-genome sequencing, biliary strictures, cholangiocarcinoma, pancreatic ductal adenocarcinoma, ERCP, copy-number alterations, tumour fraction, liquid biopsy, CA19-9, early cancer diagnosis, eGastroenterology