A statistical model, or nomogram, has been developed to predict long-term survival of patients following resection for intrahepatic cholangiocarcinoma (ICC), based on data from an international multicenter study.
Although the tool is still investigational, the developers report that patients in the top quartile of predicted survival had a median survival of 80.2 months, compared with 14.8 months for those in the lowest quartile (P = .01).
The authors identified 6 variables that were the most closely associated with survival.
The findings were published online March 5 in JAMA Surgery by Omar Hyder, MD, from Johns Hopkins University School of Medicine in Baltimore, Maryland, and colleagues.
Intrahepatic cholangiocarcinoma is the second most common primary liver malignant tumor, after hepatocellular carcinoma (HCC), and represents 10% to 20% of all primary liver malignant tumors, or about 3100 new cases every year in the US.
Yet the disease was only recently recognized as an entity distinct from HCC and provided its own unique staging system in the 7th edition of the American Joint Committee on Cancer (AJCC) Staging Manual. The previous 6 editions failed to make that distinction, said study coauthor Timothy M. Pawlik, MD, MPH, PhD, Director of the Johns Hopkins Medicine Liver Tumor Center Multi-Disciplinary Clinic, Johns Hopkins Hospital, told Medscape Medical News.
Although tumor staging is helpful, disease-specific nomograms also take into account clinical factors beyond pathology. “The nomogram will provide clinicians with a patient-specific way to predict survival following resection of ICC,” said Dr. Pawlik, who is also the John L. Cameron MD Professor of Alimentary Tract Diseases at Johns Hopkins.
Asked to comment on the study, David A. Geller, MD, Director of the UPMC Liver Cancer Center, University of Pittsburgh, in Pennsylvania, told Medscape Medical News, “developing a nomogram to predict survival after liver resection is very important,” noting that ICC incidence is increasing and overall prognosis is poor.
The majority of patients are inoperable at diagnosis, and in those who are able to undergo hepatic resection — the only chance for cure — overall 5-year survival is only 25% to 35%, added Dr. Geller, who was not involved in the study and is also the Richard L. Simmons Professor of Surgery at UPMC.
According to Dr. Geller, “This is a well-done multi-center, international study from leading centers around the world.” Once validated in external cohorts, “[the nomogram] potentially could be used to counsel patients and families about risk of recurrence, and could be one of several factors used by medical oncologists to discuss risks/benefits of adjuvant therapies.”
Dr. Pawlik said that the tool can be employed now. “Although the nomogram can be used now to help shape the discussion around patient prognosis, future validation should be performed.”
Six Variables Used to Predict Survival
Hyder and colleagues used data from a study of 514 patients who underwent resection for ICC at 13 major hepatobiliary centers in the United States, Europe, and Asia from 1990 through 2011. Approximately half (53%) of the patients were male and nearly two thirds (61%) were white. They had a mean age of 59 years.
Extended hepatectomy was performed in 39%, hemihepatectomy in 35%, and minor live resection in 26%. Median tumor size was 6.0 cm, with 56% having tumors of 5 cm or larger. Most of the patients (88%) had R0 surgical margins. Lymphadenectomy was performed in 49%, and 24% received adjuvant therapy.
Mean survival postsurgery was 39 months, with survival rates at 1, 3, and 5 years of 81%, 52%, and 40%, respectively.
From an initially selected 15 clinically relevant candidate variables, the authors identified 6 that were the most closely associated with survival. Of the 6 variables, 5 had statistically significant associations with P values < .05: age (hazard ratio [HR], 1.31); presence of multiple tumors (HR, 1.58); tumor size (HR, 1.50); lymph node metastasis (HR, 1.78); and macroscopic vascular invasion (HR, 2.1). The other variable, the presence of cirrhosis, carried a P value of .08 (HR, 1.51).
Further statistical examination of the effect of the 2 continuous variables — age and tumor size — revealed that both had a nonlinear effect on the risk for mortality. Specifically, the effect of tumor size was linear below a threshold of 7 cm, but was constant above that. Similarly, the highest risk for mortality was seen at extremes of age in both directions.
Based on the 6 variables, a nomogram was constructed with points assigned to indicate a survival prognosis. For example, lymph node metastasis was associated with 11 points, while the presence of macroscopic vascular invasion was given 15 points. Higher scores indicate worse prognosis.
The nomogram has a 69% probability of accurately predicting survival, say the study authors, referring to a calculation of the “Harrell’s C-index.”
In comparison, the probable accuracy for predicting survival in all cancer with the 7th edition of the AJCC Cancer Staging Manual is just 59%. For the 6th edition, which had staged ICC the same as HCC, the prognostic ability for ICC was 54%, the authors note.
Role of Tumor Size Reassessed
Dr. Pawlik told Medscape Medical News that his group had previously published findings showing no effect on tumor size. As a result, that variable is not included in the current AJCC ICC staging system.
The reason for the discrepancy with the current data may relate to the nonlinearity of the relationship between tumor size and prognosis, whereby it ceases to be significant above an approximate 7-cm diameter threshold. Previous studies may not have included enough patients with smaller tumors to have influenced the findings, the authors point out in their discussion.
This means that the AJCC staging will likely need to be revised to take tumor size into account as a predictor of survival, Dr. Pawlik told Medscape Medical News.
“Since the ICC staging in the current AJCC manual is the first version ever to be published, future refinements are to be expected. The nomogram is an important next step,” he said.
The authors point out that 1previous nomogram for ICC has been published, but it was based on data from just 1 center in China and evaluated age and tumor size as strictly linear variables.