Hi Christer- I guess you are asking because you want to start a discussion.

I think we need biomarkers that can be used to enrich for the intact CTCs ie cell surface markers that signal a cancerous cell regardless of origin. Not sure there are any...

We also need biomarkers that are abundant enough to be detected using available technologies. Immuno-histochemistry allows single-cell sensitivity, but only allows one marker at a time to be visualised by traditional clinical techniques, or a few markers if using fluorescence, which is not widely used in the clinic.

Then we need biomarkers specific for tissue of origin, which should be identifiable from the literature, but would need to be confirmed experimentally.

Ideally, we would have markers to distinguish between (1) a cell from a primary tumour versus a metastasis; (2) a cell from a non-aggressive, benign or or indolent growth versus a malignancy; (3) a genetically stable versus an unstable cell; (4) a drug-resistant versus drug sensitive cell...

To make life easy, all the markers would be cell-surface markers, allowing rapid diagnosis using FACS or similar, as is already happening in haematology. Unfortunately, much more is known about intracellular markers in cancer and in drug resistance, limiting analysis to immuno-histochemistry or -cytochemistry and making multiplexed analysis more difficult. Luminex and Rules-Based Medicine have resources there, but the numbers of markers that can be looked at are limited, and are not generally cancer-focussed.

Is this the kind of response you are looking for- a general one? Or did you want specific ideas?

Paul

It seems to me it would be good to first inventory the existing state of the clinical art. That would be clinical, pathological, markers to establish the origin of "cancers of unknown origin". Then appraise to what extent that fulfils the diagnostic needs, and where the unfilled needs are for each major cancer type may be. Then map out the road ahead, whether it is collecting and clinically validating already known determination or differentiation markers or establishing the need for new ones.

Christer and Paul,

I believe that the most important first thing is to define the goal - is it just an academic exercise or assigning CTCs to a specific tumor type has practical applications? If practical, then what exactly? IMHO, diagnostic value of CTCs is limited, and so is predictive value (no CTCs in early stages, too few CTCs to draw conclusions even once they can be detected). We are left with prognostic value and monitoring. With regards to prognostic value I have grave doubts, primarily because once the primary is defined, it is much easier to draw prognostic conclusions from its analysis. Now monitoring is another issue, especially for non-resectable tumors (these are already identified otherwise we don't know if they are resectable, right?). Again, I have doubts that a small number of heterogeneous CTCs will provide a definitive value.

For practical applications a biomarker has to be objective, the test should utilize accessible tissue, and be simple. Easily accessible are blood, saliva, and urine (I am not talking about digestive diseases, where feces can be used as well as blood, etc). Upcoming are markers based on VOCs from breath. In terms of physical basis biomarkers can be (a) proteins - antigens, autoantibodies, modified proteins, peptides; (b) nucleic acids - RNA, DNA, and their modifications and subtypes; (c) lipids; (d) metabolites. My hat is in the ring for DNA modified by methylation - we can (at this moment) differentiate different diseases, benign and malignant disease, maiignant and inflammatory, early and late stages, and effects of treatment. DNA is cell-free circulating DNA from blood, which carries a lot of tumor-related methylation features. Considering that a very high resolution mapping is feasible, methylation-based biomarkers seem to fit the bill.

Great points Victor! How do you respond to those who say that the circulating tumor DNA is diluted (and so may in some critical situations be undetectable) and in any case mixed with more abundant normal or non-tumor DNA? And how are you to connect a given DNA molecule positively, rather than statistically, to a cancer cell origin if you don't have a diagnostic mutation or rearrangement in the particular fragment under study? Isn't it good to be able to make the link to a cell? And how many clones might there be in the circulation at any given point in time? Maybe you don't need that many cells if you have an accurate assay?

Christer, good counterpoints! I must confess that I am not a theoretician but rather an experimentator, so when the Q about heterogeneity came up, we did a study (Melnikov, JMolDx, 2008) where we have shown that even in very heterogeneous media (gross sections of tumors) we see specific methylation patterns (it would have been nice to do LCM and compare, but at that time we did not have the ability to work with ultra-small samples). Undetectable cfcDNA is a serious concern, but - again, based on experience - less than 1-2% of samples do not produce at least 0.5 ng/ml and that is usually due to the operator's error. cfcDNA is by definition heterogeneous and contains pretty much everything that can get into blood; our approach is that we discard all features common to the two comparison groups and concentrate on those that are different. My colleague has a nice saying that comes along - ancient mariners used the Polar star for navigation for many years before formal astronomy could catch up and explain how and why. If we assume that our knowledge is by definition very limited, then we have to rely on simple correlations that only later will be mechanistically explained. Only then we will be able to predict associations of methylation in fragment A with disease development (BTW, no mutations or rearrangements here ).

Link to the cell is interesting but again - why do we need it? It is well established that stroma of breast cancer has changed as well, so what cells do you want to connect to and what is the value of this connection? In a way we have to determine a systemic response of the organism to tumor (or disease in general) emergence, rather than its link to a specific cell.

The same (why do we need it?) Q relates to the number of CTC clones.......

Please refer to our journal regarding this matter. In this article discussed the most common alterations genetic changes that we are able to detected in biological fluids such as blood, urea, serum and ....... This article published in Avicenna J Med Biotech 2012; 4(1): 3-13 with title “Non-Invasive Detection of Esophageal Cancer using Genetic Changes in Circulating Cell-Free DNA.”

http://www.ajmb.org/En/Issue.aspx?id=12

Thanks Saeid! Got it!

Just found this question so I'm sorry to be late. I believe you would need an array of biomarkers to determine site of origin if this were unknown. For instance, Prostate Specific Membrane Antigen (PSMA) is over-expressed in prostate cancer and HER2 is an obvious target for breast cancer. These are examples of biomarkers that are more specific for single cancers. There are also other biomarkers such as EGFR, MUC-1, and EpCAM that are found in various cancers and circulating tumor cells. So a panel of markers is more likely needed.

Article Prostate Cancer Targeting Motifs: Expression of αvβ3, Neurot...