Tumors are com­plex sys­tems of cells, only some of which may be can­cerous. Also, two sam­ples from dif­ferent areas of a single tumor are rarely iden­tical. To gather impor­tant infor­ma­tion about tumors, researchers must ana­lyze very small sam­ples because they are more likely homoge­nous — enriched for either normal cells or can­cerous cells.

Barry Karger — the James L. Waters Chair in Ana­lyt­ical Chem­istry in Northeastern’s Col­lege of Sci­ence — and col­leagues at the Mass­a­chu­setts Gen­eral Hos­pital Cancer Center have iden­ti­fied a net­work of genetic markers capable of pre­dicting the relapse of estrogen-​​receptor pos­i­tive, or ER+, breast cancer. The Susan G. Komen Breast Cancer Foun­da­tion funded the research.

We have genomics, pro­teomics, metabolomics,” said Karger, who also directs the Bar­nett Insti­tute of Chem­ical and Bio­log­ical Analysis. Genomics is the study of all of a cell’s genes; pro­teomics and metabolomics examine a cell’s pro­teins and metabo­lites, respec­tively. “How do you put these dif­ferent ‘omics’ together to gen­erate meaning?”

Karger and his long-​​time col­lab­o­rator, breast-​​cancer pathol­o­gist Dennis Sgroi, had asked this ques­tion for a while. But they could not fully address the problem until the duo teamed up with bioin­for­matics spe­cialist Marcin Imielinski, whose com­pu­ta­tional know-​​how proved crucial.

In pre­vious work, Sgroi had iso­lated highly enriched cancer cell sam­ples by per­forming microdis­sec­tion on ER+ tumor tis­sues. Gene-​​expression analyses of these sam­ples were com­pared with those of benign cells to iden­tify a series of genes asso­ci­ated with malignancy.

We’d been doing gene-​​expression analyses for years,” said Sgroi. “But anal­o­gous pro­teomic inter­ro­ga­tion of lim­iting amounts of clin­ical sam­ples was a sig­nif­i­cant challenge. ”

Karger pointed out that gene expres­sion is only one part of the story: “The pro­tein is where the actual func­tion is,” he explained. “You express a gene that then pro­duces a pro­tein, but the pro­tein does the work.”

If the researchers could per­form pro­teomic analyses on small sam­ples, they could cross-​​reference those data with the gene-​​expression data. But it was simply not pos­sible until last year, when Karger pre­sented a new ana­lyt­ical method, which does just that.

By com­bining genomics and pro­teomics, the team now had the nec­es­sary tools to gen­erate a net­work of poten­tially rel­e­vant genes that would be more exten­sive than net­works gen­er­ated from either genomics or pro­teomics alone.

To test the sig­nif­i­cance of the hybrid net­work, the researchers applied it to the gene pro­files of more than 600 ER+ patients. The net­work had a sta­tis­ti­cally sig­nif­i­cant ability to pre­dict which patients relapsed out of remis­sion, making it a prog­nostic bio­marker for ER+ breast cancer.

Many breast cancer bio­markers already exist — but few are very robust, said Sgori. The novel approach gives the work more impor­tance, he explained: “We feel it serves as an impor­tant proof-​​of-​​concept and step­ping stone for future studies.”