Predicting Pancreatic Cancer Survival

Cancer in all its inglorious forms remains a killer but there are signs that the huge investment in cancer research, improvements in cancer care and, possibly, lifestyle changes, are having an effect. In the USA, the National Cancer Institute announced that the rates of new diagnoses fell about 1% per annum from 1999-2006 while death rates from all cancers dropped about 1.6% per annum from 2001-2006.

These falls were largely to lung, prostate, and colorectal cancers in men and breast and colorectal cancer in women. For other individual cancers, the picture is not so bright. For instance, increased incidences of liver and kidney cancer in men and lung, bladder and pancreatic cancer in women were reported.

In fact, invasive ductal pancreatic cancer is the fourth highest cancer killer in the USA with more than 33,000 deaths each year and the fifth largest in Japan, at more than 23,000 deaths annually. It is not helped by the fact the most patients have undergone distant metastasis, where the cancer has spread to other organs or distant lymph nodes, at their first clinical presentation.

The standard treatment is chemotherapy with the drug gemcitabine, a nucleoside analogue that aims to stop the uncontrolled cell division occurring in the tumour by replacing the nucleic acid cytidine and preventing DNA replication. The drug has improved the median survival times of pancreatic cancer patients, but their responses are variable and a significant number of patients receive little benefit. In some cases, cancer patients have longer survival by changing the type of drug treatment, or even by withdrawing it completely.

It follows that a method for classifying patients into groups likely or not to respond to gemcitabine treatment would be of great help to medical staff, so that the treatment regime could be tailored for each individual.

This goal has been targeted by scientists in Japan, in an 18-person, 6-organisation proteomics study reported in Molecular and Cellular Proteomics by Junichi Matsubara from the National Cancer Centre Research Institute in Tokyo. They compared the plasma proteomes of patients with advanced pancreatic cancer who had received the same gemcitabine therapy, 29 dying within 100 days and 31 surviving more than 400 days.

Proteins in the plasma were digested with trypsin and the peptides were analysed in triplicate by LC-tandem MS with electrospray ionisation. The peaks were detected, normalised and quantified by an in-house software package (2DICAL) which quantifies protein content in an unlimited number of samples without resorting to isotope labelling techniques.

The mean intensities of the triplicates differed significantly between the short-term and long-term survival patients for 637 peptides. From these, the team identified the peaks that were most increased in short-term survivors proteins as alpha1-antitrypsin and alpha1-antichymotrypsin. The differential expression of both proteins was confirmed by SDS-PAGE and immunoblotting.

Although they appeared to correlate with survival, the levels of the two proteins did not correlate with the tumour response.

Their potential as biomarkers was tested on the plasma or serum of 304 patients with advanced pancreatic cancer who had not yet begun gemcitabine therapy. This time, reverse-phase protein microarrays were set up using a series of antibodies, with subsequent staining for fluorescence measurement. There were no differences in the results from plasma and serum.

The levels of alpha1-antitrypsin and alpha1-antichymotrypsin were not mutually correlated. However, a statistical analysis revealed that each protein had a significant correlation with overall survival. Patients with high levels of alpha1-antitrypsin had shorter survival times than patients with low levels: 176-219 days (median 210) compared with 271-439 days (median 327).

The same pattern was observed for alpha1-antichymotrypsin, with high levels signifying a shorter survival period: 193-235 days (median 211) versus 255-416 days (median 327).

Other values such as leukocyte count and alkaline phosphatase also correlated with survival status for patients on gemcitabine. However, published work revealed that none of these factors alone was sufficiently accurate.

Using a multivariate predictive model, the researchers found that a combination of alpha1-antitrypsin, alkaline phosphatase, leukocyte count and the Eastern Cooperative Oncology group (ECOG) performance status performed the best. The exclusion of alpha1-antitrypsin from the model "significantly compromised" its performance.

These findings were used to devise a scoring system (nomogram) that incorporated the four values into a single score that estimated the survival outcome. It proved to be highly accurate, giving survival times of 123-187 days (median 150) for patients with a score >94 and 255-328 days (median 282) for scores <94.

The clinical significance of alpha1-antitrypsin itself remains unclear. Its main function is to inhibit the protease activity of neutrophil elastase but its absolute levels varied widely from patient to patient and did not appear to correlate with the efficacy of gemcitabine treatment. This suggests that it might reflect the natural course of pancreatic cancer, regardless of treatment.

Nevertheless, alpha1-antitrypsin is an accurate biomarker of survival times. The nomogram will be useful for prognosis, earmarking patients for whom gemcitabine will not increase survival times. This raises the possibility of tailoring the treatment of pancreatic cancer to individuals, especially when alternative drug therapies are launched in the future.

Related links:

• Molecular and Cellular Proteomics 2010 (Article in Press): "Survival prediction for pancreatic cancer patients receiving gemcitabine treatment"

Article by Steve Down

Source: Proteomics and Genomics