CDx expert: “Genomics might only be the tip of the iceberg.”

January 19, 2023
20 min read

It will soon be 25 years since the first CDx assay came to the market paving the way for the start of the era of personalized treatment. We have discussed the last 25 years of CDx, the achievements within the field and other important topics with Dr. Jan Trøst Jørgensen of Dx-Rx Institute in Denmark.

Good to talk to you today, Jan, to introduce you to our readers, you have an extensive academic and industrial career in clinical and drug-diagnostic development and worked for or cooperated with Novo Nordisk, Novartis, Dako/Agilent Technologies amongst others. You are also a reviewer and editor for several journals and a director of the Dx-Rx Institute in Denmark. Maybe you can explain in a little bit more detail how intensively you have been involved with the field?

I have a 40-year career in the pharma and the diagnostic industry, mainly within clinical development, and for the past more than 20 years, I have worked with companion diagnostics. My academic background is an M.Sc. in pharmaceutical sciences, and a Ph.D. in clinical pharmacy. I have worked a lot with pharmacotherapy, and here we have for decades discussed the concept of rational use of a drug or rational pharmacotherapy - the right drug for the right patients at the right time, and so forth, doing all the right things. This was a concept I was introduced to through my basic pharmacology lectures at the university more than 40 years ago. However, at that time, it was easier said than done because of our limited knowledge of molecular pathology and the drug mechanism of action, but things have changed. Progress in molecular medicine has provided us with a better understanding of these aspects, which has positively impacted pharmacotherapy. Especially within disease areas such as oncology and hematology, great progress has been made with the introduction of several targeted drugs guided by predictive biomarkers, or what we today call companion diagnostics.

So, from your perspective, how do you define companion diagnostics in a few words?

If we take on the glasses of the regulator (FDA and EMA), a companion diagnostic is defined as a predictive biomarker assay that provides information that is essential for the safe and effective use of a corresponding drug. By using a companion diagnostic for therapy selection, you move away from the “trial and error” approach and the prescribing process becomes more rational. In oncology, many targeted cancer drugs have been developed for a very small subset of patients, and it would be unethical and financially irresponsible to administer these drugs to patients without using a companion diagnostic assay for treatment selection. For instance, in patients with NSCLC, the prevalence of ROS1-positive tumors is approximately 1%, which qualifies them for treatment with crizotinib or entrectinib. Therefore, the administration of these drugs to patients without prior testing using a companion diagnostic assay has no meaning.      

There are different players involved in the field. From the history what would you say are the key industry players, and do you see this changing soon in terms of the market dynamics?

Dako/Agilent Technologies was the first company to develop a companion diagnostic assay, which was approved by the FDA. HercepTest was developed for trastuzumab (Herceptin) for the treatment of HER2-positive metastatic breast cancer patients. Therefore, Dako/Agilent Technologies is definitively a diagnostic company with a lot of experience within this field, and this September, it is 25 years since trastuzumab and the HercepTest were simultaneously approved by the FDA. In fact, two years ago, we published an article in Frontiers of Oncology on the development of this assay ( Besides Dako/Agilent Technologies, several other companies have experience with companion diagnostics, such as Roche Diagnostics, Qiagen, Abbott, ThermoFisher/Life Technologies, and Foundation Medicine.

Do you believe that this kind of dominance of, let's say Roche and Agilent, in the space will continue also in the future? Because the oncology development is now really across the board. Almost every mid-sized pharma company has some kind of oncology pipeline. Do you think that it's going to shift in terms of market dominance?

There are a few major players and I have mentioned Agilent Technologies, which bought Dako ten years ago, and companies such as Roche Diagnostics, Qiagen, ThermoFisher, and a few more. If you are a pharmaceutical or biotech company and want to find a partner for developing a companion diagnostic, this partner must have a strong position within its field and an organization that can manage regulatory submissions. Furthermore, they must have a worldwide presence to manage the sales and distribution of the assays. This is why these major diagnostic companies have advantages over smaller ones and often will be their preferred partners.

There has been a recent breakthrough in breast cancer therapy with the expanded stratification of HER2. There is a new kind of therapeutic subgroup which has been presented at ASCO this year, HER2-low, which accounts for about 50% to 60% of cases that were not really specifically clinically actionable, and now with the approval of this antibody-drug conjugate have become nicely actionable. How often do you see such a dramatic shift in the stratification provided by a CDx biomarker?

The results from the DESTINY-Breast04 study presented at the last year’s ASCO meeting must be regarded as a major breakthrough. Trastuzumab deruxtecan (Enhertu) is an antibody-drug conjugate developed by Daiichi Sankyo and AstraZeneca. To date, metastatic breast cancer patients with HER2-low tumors (IHC2+ or IHC1+ and ISH negative) have most often been offered treatment with standard chemotherapy, and it now seems that this relatively large group of patients has received an additional option. The DESTINY-Breast04 study in patients with HER2-Low tumors showed a significantly longer progression-free survival and overall survival with trastuzumab deruxtecan than with standard chemotherapy. In Europe, trastuzumab deruxtecan has so far only been approved for the treatment of patients with metastatic HER2-positive breast cancer who have received one or more prior anti-HER2-based regimens. In the USA, the HER2-low indication was approved in August last year. One thing is the approval by EMA of the HER2-low indication, another thing is reimbursement of this relatively expensive drug, so my impression is that we still have to wait a while before it can be offered to breast cancer patients. Another challenge is the HER2 assays used to detect the HER2-low status. A recent study published in JAMA Oncology showed that the concordance for the HER2 scores of IHC0 and IHC1+ was quite poor, which raises the concern that some patients will be erroneously misclassified with regard to treatment with trastuzumab deruxtecan.

Luckily, we have received the news a couple of months ago that Roche developed the CDx assay specifically for the assessment of HER2-low. We don't know how it's been performing on the market yet, it's been approved, but we don't see the vigilance data. But it's going to be interesting to look at for sure.

Yes, it is important to see how this assay will perform and if it will be helpful to stratify HER2-low breast cancer patients for treatment with trastuzumab deruxtecan. As far as I understand, this assay is only available on the market in the USA. 

Historically companion diagnostics have been very closely tied to oncology but we now see also drugs approved in different indications where they are approved alongside a CDx device. We see some drugs in hematology and obesity. What do you predict for CDx in terms of spread across indications? Where do you see the next steps in terms of the disease areas?

For years, drugs for the treatment of hematological diseases have been guided by companion diagnostic assays, which is not a new development. In general, when we are talking about companion diagnostics today it is mainly related to oncological and hematological drugs. When setmelanotide (Imcivree) was recently approved by the FDA for chronic weight management in patients with monogenic or syndromic obesity, it was combined with a companion diagnostic assay for the detection of the specific genetic aberrations. However, this is a rather rare disease, so it is a very small number of patients who will benefit from this treatment. Besides setmelanotide, the FDA has only approved one additional drug with a companion diagnostic outside the hematological and oncological disease area, deferasirox (Exjade), which is used to treat chronic iron overload in patients with thalassemia. This companion diagnostic assay is based on an imaging biomarker, which is a bit unusual. However, in the years to come, we will see companion diagnostics being developed for disease areas outside oncology and hematology. This could be within autoimmune and cardiovascular diseases as well as different metabolic disorders. The use of companion diagnostics is a means to individualize pharmacotherapy - the right drug for the right patients – or what we today call personalized or precision medicine. Molecular medicine has provided us with new tools, so we can slowly start to practice this concept. Another aspect is the high prices of new drugs, and here, the payers require value for money. This means that high efficacy and companion diagnostics can help to achieve this goal by preselecting a group of likely responding patients. That high response rates can be achieved by combining drug and diagnostic have been shown in different publications. Based on the regulatory documentation submitted to the FDA, I described the response rates for a number of targeted cancer drugs guided by a companion diagnostic assay in an article in Trends in Molecular Medicine. The response rates for the listed drugs ranged from 34% to 80%. Recently, we have seen even higher figures with the tumor agnostic drug larotrectinib in patients with NTRK fusion-positive tumors. In a study published in Lancet Oncology, the response rate across different tumor types was 79%, and in a group of patients with soft tissue sarcoma, the response rate was close to 90%.

Selecting the right biomarker, whether in oncology or other indications, is quite important. When would you personally in the development of CDx assay consider the biomarker to be validated? What kind of sample size or what kind of conditions would warrant the validation of the biomarker as a suitable biomarker for a particular drug?

Before you start to use a companion diagnostic assay on patient samples/specimens in a pivotal clinical trial, you will need to ensure that it is sufficiently analytically validated, so you can rely on the results for patient stratification. You will need to do all the things you normally will do when you perform biomarker validation and also make sure that you fulfill the requirement of the regulators, which, among other things, will address the preanalytical factors, sensitivity, specificity, robustness, repeatability, and reproducibility. In a clinical trial, we need to document the predictive value of the companion diagnostic assay by showing a clear link between the assay result and the outcome following treatment with the investigational drug. Thus far, the FDA has been at the forefront of regulatory requirements concerning the analytical and clinical validation of companion diagnostic assays. However, with the new IVD regulation (EU 2017/746), which came into force last year in May, we now face changes in Europe. Self-certification by diagnostic companies or commercial laboratories is no longer an option. All new companion diagnostic assays introduced in the EU must be CE-marked according to the new IVD regulation, which requires a regulatory review by an independent notified body and, in many situations, the national competent authorities or EMA.

Yes, CDx devices are rather high-risk according to the EU and it seems to put a lot of pressure on the manufacturers. We've seen the CDx in oncology within a spectrum of different platforms. You can look at the gene variant in the host itself, you can look at the biopsy of the tumor tissue. So, we were able to see immunohistochemical assays like the original one for HER2, all the way to the newest NGS panels. What do you expect to see in terms of the development of the technology platforms? Can we expect an increase in the NGS-based CDx solutions? Are they size-wise going to stay at the level of what the PCR can provide, so just a few genes? Which kind of technology platform is going to play a major role in the near future?

Both FDA and EMA classify companion diagnostics as high-risk devices. According to the new IVD regulation, these assays are classified as class C, which means high individual risk and low public health risk, underlining the critical role they play in the treatment of individual patients. Over the past 25 years, there has been an interesting development in the analytical platforms used for companion diagnostics. Starting with immunohistochemistry closely followed by different types of in-situ hybridization assays. Ten-fifteen years ago we saw the PCR assays being introduced and then in 2017 the FDA approved the first companion diagnostic assay based on NGS. NGS will play an increasing role and in the USA five or six NGS assays have already obtained FDA approval. Initially, these assays covered a relatively small number of drugs and these panels are now expanding. An example is the FoundationOne CDx assay, which currently covers 28 different drugs or drug combinations. When it comes to the analytical and clinical validation of different types of assays, it is more or less the same thing you have to do as previously discussed, with some variation depending on the technology. An analytical platform not yet approved as a companion diagnostic assay is mass spectrometry, which may only be a question of time. In the future, proteomics will play a much greater role, as we need to bridge the gap between genotypes and phenotypes. There are approximately 22,000 coding genes, and according to an estimate from the National Cancer Institute in the US, there may be up to 1 million different proteins. Genomics might only be the tip of the iceberg, and to improve the treatment of cancer patients, a much better understanding of the link between genotype and phenotype is needed.     

Not counting the genes that only express to the mRNA level...

You are right; we should not forget the transcriptome. The National Cancer Institute estimates that it covers approximately 100,000 mRNA molecules. We have a lot to learn how these different molecules can help us predict the response to pharmacotherapy, and when it comes to companion diagnostics, we will likely have to integrate different analytical platforms, which also include older technologies such as IHC in a kind of “composite biomarker”.

It's good to hear that immunohistochemistry is not so old-fashioned after all.

The combination of IHC with image analysis and artificial intelligence will turn it into a strong technology that will definitely survive in the future. I think we will soon pass the situation where pathologists do not have to evaluate every single slide manually because of the new digital image analysis tools.

There are successful start-ups, also here in our region, analyzing different kinds of pathology slides with various types of tissues and they do have a pretty good sensitivity and also quite a good specificity. It's between 80 to 90%, so there are some specific nuances but it's quite good already.

These technologies are superior in terms of their reproducibility. Image algorithms can analyze tissue slides faster and more reproducibly than the human eye. This technology will keep immunohistochemistry alive and very likely in situ hybridization assays such as FISH. We recently published a paper in Cancers on companion diagnostics for MET-targeted therapy (, in which amplification of the MET gene seems to play an important role as a resistance mechanism to tyrosine kinase inhibitors for the treatment of non-small cell lung cancer. However, it seems that NGS is not able to distinguish between true MET amplifications and polysomy, which is quite important because it is the true amplification that drives the process. The analysis of FISH slides is complicated, and counting the signals from genes and centromeres is time-consuming; however, this may change by using image analysis and artificial intelligence.

Let's move from technology to the reimbursement landscape and regulatory affairs. We have already mentioned the new IVDR directive. We know that CDx devices have been classified as high-risk C and now EMA is also responsible for issuing an opinion on the suitability of the drug-device pair. What do you see in terms of the development of the CDx landscape in Europe in comparison to, let's say, the US as a result of this new development?

For me, the new IVD regulation is a question regarding patient safety, and I hope that it will lead to more accurate and precise companion diagnostic assays. False-negative and false-positive test results can have severe negative consequences for individual patients as they might end up receiving the wrong treatment. Another consequence of the new IVD regulation is that laboratory-developed companion diagnostic assays more or less will disappear when the regulation is fully implemented.  A negative consequence of the new IVD regulation is that it favors larger companies, as the smaller ones might have difficulties in living up to the new requirements concerning the increased level of analytical and especially clinical documentation.

What about the US?

In the USA, the FDA has regulated IVDs, including companion diagnostics, for decades but despite of this a large part of companion diagnostics used are laboratory-developed tests (LDT), which not are regulated by the FDA. For several years, it has been heavily discussed if the FDA should regulate LDT. The VALID Act was proposed to the US Congress, which would give the FDA more authority in relation to LDT, but it has still not been passed. Although this act has not been passed, the discussion is not over as this is again a question about patient safety. I am nearly sure that we see some kind of FDA oversight when it comes to LDT in the USA.  Companion diagnostic assays must be of high quality with respect to their performance; so, clinicians can rely on test results when selecting therapy for individual patients.

How do you believe are the reimbursement systems, and payers - meaning the insurance companies, and different regulatory and assessment bodies in the EU prepared for the increasing adoption of CDx solutions? A report from Nature published last year said 66% of all drugs approved by the FDA in 2021 had genomic evidence supporting the choice of the drug target. So presumably we can expect many new suitable biomarkers and consequently CDx devices to emerge soon.

Undoubtedly, predictive biomarkers will play a much more prominent role in the future, and payers will need to find an appropriate solution for reimbursement issues. When you mention that 66% of all drugs approved by the FDA in 2021 had genomic evidence supporting the choice of drug target, I am sure that we are not talking about companion diagnostics. As far as I remember, there were six or seven drugs approved by the FDA with a companion diagnostic assay linked to their use in 2021. In a recent editorial in Expert Review of Molecular Diagnostics, we discussed predictive biomarkers with a focus on companion diagnostics and pharmacogenetic assays ( While companion diagnostics are most often linked directly or indirectly to a specific drug mechanism of action, pharmacogenetic biomarkers are frequently linked to the metabolism of drugs, such as cytochrome P450 sub enzymes. Therefore, for most of these approved drugs (66%), we are likely talking about different pharmacogenetic aspects. 

CDx devices, before they are approved for market use, are typically used in the stratification of patients during clinical trials to optimize enrollment. What do you see as the most common issues associated with the use of these assays for investigative use only (IUO) in clinical trials?

As briefly discussed earlier, these assays must be sufficiently analytically validated before starting to use them in a clinical trial for patient stratification. When conducting clinical trials where a companion diagnostic is included, you will often seek confirmation of a biomarker hypothesis. Therefore, if you should be able to show a link between the assay result and clinical outcome following treatment with the investigational drug, you need to be able to rely on the test results; otherwise, such a clinical trial gives no meaning. This is why the analytical validation of the assay is critical before starting a pivotal clinical trial. Furthermore, the use of companion diagnostics in clinical drug development is a very powerful tool. The enrichment trial design, in which patients are enrolled based on their biomarker status, has shown that drug efficacy can be documented in a relatively small number of patients. A recently published paper in Cancer Treatment and Research Communications ( showed that for a relatively large number of drugs, populations of fewer than 200 patients, and sometimes below 100, were sufficient to document clinical efficacy. All these drugs have obtained FDA approval within the past 5-10 years based on such type of documentation.

You also published an article on this recently - on missing companion diagnostics for new drugs and indications, which was very interesting. What are the most common reasons why the assays, when they are already used in the IUO setting and they've been analytically validated, do not make it through to the approval with their respective drugs and stay as unapproved assays?

Yes, it is also a bit of mystery to me. In an article published in Journal Clinical Oncology Precision Oncology last year (, I discussed a trend that we have observed for the past 5-6 years, where new drugs and new indications were approved by the FDA without a companion diagnostic, despite the use of a predictive biomarker assay for patient selection during clinical development. The drug-diagnostic codevelopment model is a very powerful tool, but it also means that pharmaceutical and biotech companies should aim for simultaneous approval of drugs and diagnostics. In my opinion, it gives very little meaning to bring a targeted drug to the market without a validated assay that can select the right patient population for treatment. There are likely several explanations for this situation, and regulatory authorities are also partly to blame. They should be stricter concerning the requirement, so an analytical and clinically validated assay is available at the same time as the drug. However, the main reason for this situation is likely to be found among pharmaceutical and biotech companies that conduct the clinical development of the drugs. The only major reason I see here is poor project planning. Their main focus is on the drug and not the companion diagnostics, and they often underestimate the time and resources needed for diagnostic development. For several of the drugs and indications mentioned in my paper, we often saw that several different locally developed assays were used for patient selection in the trials, and if you know a bit about laboratory-to-laboratory variation, this is not good enough. One question you could raise here is how confident we can be about the patient population enrolled in these clinical trials. I hope that this will improve in the future because this is again a question about patient safety, which was also the reason why I wrote the article.   

It is very hard to understand for me as well, especially if it's otherwise a perfectly usable device.

However, it was probably not a perfectly usable device. When you do a drug-diagnostic codevelopment project, you must start as early as possible to develop a companion diagnostic assay. It is strongly recommended to have a more or less final analytical validated assay ready when you start your pivotal clinical trial; otherwise, you have to perform a subsequent bridging study to bridge between the non-analytical validated assay and the validated one, which is both costly and time-consuming.

When you look at the CDx field and the CDx devices holistically, what do you see as the most prominent issues? Is it bad planning in terms of validation, so you come up with a device that doesn't have sufficient analytical power in the end? Regulatory burden? Is it just simply that the pharma companies which are still the main CDx (co)developers lack expertise? We have seen in some cases in the past, e.g., the Exubera case, that pharma is naturally good at drugs but not necessarily well-versed in fields outside of drugs? 

If we go 5-10 years back in time, the knowledge within pharma and biotech was limited, but today most of them have specific departments that specialize in companion diagnostic and predictive biomarkers to support their drug development activities. Missing or insufficient assay validation is often a result of poor project management. Another important aspect to consider when working with companion diagnostics is laboratory-to-laboratory variation as I briefly mentioned before. It is well known from IHC and ISH assays, and great efforts have been made to improve the situation by organizations such as Nordic QC and UK NEQAS, which perform external quality assessments through proficiency testing. Unfortunately, this problem is not linked to IHC and ISH assays only. Recently, a study conducted in the USA with NGS testing at 21 different clinical laboratories and published in American Journal of Clinical Pathology, showed that variability in accuracy across laboratories resulted in the selection of different patient populations for targeted therapy. There is a lot of work to be done to improve the quality of companion diagnostic testing to minimize laboratory-to-laboratory variation. Of course, it should be so that patients tested in Slovakia, Germany, or Denmark with the same molecular tumor aberration should have the same diagnosis and treatment independent of the laboratory where the patient specimen was tested.         

It's been an enriching discussion. Thank you for the experiences in this field that you could share with us.

Let us talk about we at ambiom can be helpful to your CDx efforts at And do not forget to subscribe to our newsletter for more interesting content.

The links to the publications inserted in the interview are all open-access articles and can be downloaded free of charge from the homepages of the different scientific journals.  


No comments.


ambiom s.r.o.
Karpatské námestie 7770/10A
831 06 Bratislava
IČO: 54219035

+421 948 652 276



Back to top Arrow