OXFORD, England -- A technique that can detect 10 malignantcells among 10 million healthy ones is about to go into clinicalpractice here to aid cancer diagnosis and prognosis. It's basedon the little-known finding that the single gene encoding CD44can put out as many as 32 different products, some of whichreact characteristically with tumor cells.

CD44 is a protein-like molecule, that antibodies on the surfaceof white blood cells recognize as antigens. (CD stands for clusterdifferentiation.)

Diagnostic pathologist David Tarin of Oxford University's JohnRadcliffe Hospital, described the CD44 gene's "alternativesplicing" versatility in last week's Lancet. Besides pinpointingmalignant cells, his paper stated, the technique can predictwhich ones are potentially metastatic.

As Tarin explained to BioWorld, the CD44 gene has five exons-- coding regions -- separated by non-coding introns. When thelatter are cut out, the exons can assemble in varying order, say,1-2-3-4-5, or 1-3-5 or 2-4-5, and so on, up to 32 permutationsin all. Each variant sequence presumably encodes a specificvariant gene product, depending on the needs of a givenleukocyte population at a given moment.

"It was previously unknown that the CD44 gene can be splicedlike that," Tarin said. "A lot of people are working on CD44 andon tumor metastasis, but as far as we know, there's nobodyworking on the clinical diagnostic applications of this. We're thefirst." (But see sidebar.)

Tarin and his research associate, Yasuhiro Matsumura,compared tumor tissues freshly removed from 34 breast andcolon cancer patients with 11 similar but non-malignantsamples. They extracted messenger RNA from these cells anddeleted their non-coding intron regions to constructcomplementary DNA sequences, which they then amplified byPCR (polymerase chain reaction).

Cancerous cells, Tarin reported, but not healthy ones, "grosslyoverproduced" each of nine or more alternatively spliced largemolecular isoforms of the alternatively spliced, multipurposeCD44 gene. Comparing this excessive output with the normaltissue, "it is not difficult to see which is tumor and which is not.A non-expert could tell the difference."

Moreover, the Oxford team was able to differentiate 23metastatic tumors from eight non-metastatic ones.

The university's technology transfer arm, Oxford Innovations,applied to patent the process some time ago. Until lastSaturday's Lancet, Tarin said, "We have not spoken to anybodyabout this because we felt we had discovered something thatwas really staggering in its implications, and we wanted to getpatent protection first."

Tarin told Lancet that his system can detect down to 10 tumorcells among the 10 million cells in one ml. of blood, adding that"if this sensitivity were to be achievable with sputum, urine,stools, blood, cervical smears or other bodily fluids naturallyseeded with small numbers of tumor cells," the method "wouldhave wide applicability to cancer prevention and screening."

Tarin told BioWorld that since submitting the paper Lancetpublished last week, "we have further data, which indicate thatyou can actually detect small numbers of cells naturally shedinto bodily fluids." He plans to publish that finding soon.

Meanwhile, "we now intend to deploy it in clinical practice, ondifferent types of common cancer."


HOUSTON -- Late last May, unbeknownst to the OxfordUniversity group (see preceding story), Kenneth K. Tanabe ofM.D. Anderson Cancer Center here reported to the 83rd annualmeeting of the American Association for Cancer Research(AACR) on "CD alternative splicing in human tumors."

"Cells overexpressing specific CD44 isoforms," Tanabe told hisaudience, "display increased tumorigenic and metastaticpotential." Like David Tarin of Oxford University, Tanabeexamined CD44 expression in fresh human colon cancers, usingPCR (polymerase chain reaction), and found overexpression ofisoforms produced by mRNA alternative splicing.

Tanabe told BioWorld that Tarin's paper in Lancet "providesexciting data, which agree with our own results," but hedemurs that "the Oxford detection could be applied to clinicalscreening more readily, had the authors identified specificCD44 alternative splice variants, which were overexpressed inthe malignant tissue."

Tanabe, on the other hand, informed the AACR meeting herelast May of his finding that "primary colon carcinomas, andliver and lymph-node metastases overexpressed a specificepithelial CD44 isoform, know as CD44R1, whereas pairednormal mucosa did not." -- David N. Leff110592CD44

-- David N. Leff Science Editor

(c) 1997 American Health Consultants. All rights reserved.