Testosterone Measurement – Which Test and When? - Martin Gleave

September 25, 2019

Martin Gleave presents on the topic of testosterone measurement and its various nuances in the context of advanced prostate cancer at the Advanced Prostate Cancer Consensus Conference (APCCC 2019). Testosterone measurement is important as androgens are the drivers of activity of the androgen receptor. Dr. Gleave supports the periodic assessment of testosterone levels to ensure castrate levels, with confirmation of lab assays at low testosterone levels and suggests more study is needed for the role of testosterone and other serum androgens as prognostic and predictive biomarkers in metastatic castration-resistant prostate cancer (mCRPC). In conclusion, lower testosterone levels correlate with better freedom from CRPC.


Martin Gleave, CM, MD, FRCSC, FACS, Executive Director, Vancouver Prostate Centre, Vancouver, Canada.

Read the Full Video Transcript

Martin Gleave: Well good morning. I'd like to thank Silke and Aurelius for the kind invitation to come to this wonderful and unique meeting. I was asked to talk on testosterone measurement, which tests and when, and first I thought, "Oh boy, this is just kind of standard endocrinology. What about all the wonderful molecular biology, genomics and imaging that we could talk on?" and then I was reminded by Chris Evans that this is probably one of the most awarded molecules that we'll be speaking about today because it's been awarded four Nobel prizes, first by a German and a Swiss for actually discovering or synthesizing testosterone.

Second, a Canadian born urologist at the University of Chicago who was awarded the Nobel Prize for showing the castration prolonged disease survival and then two related Nobel Prizes to Andrew Schally and Guillemin for identifying LHRH, and at the same time a Ros Yalow for developing the RIA to detect testosterone, insulin and other peptide markers.

So again, a proud history for testosterone and it's still important today as to why we measure it. We know that androgens drive AR activity across its entire disease course. That maintenance of castrate T is required in all our guidelines. This is key for both intermittent and continuous strategies and by consensus T levels of less than 50 or 1.7 nM is defined as castrate and lower levels of serum androgens are the key therapeutic goals and now we realize, of course, with combination regimens that this is best achieved by ADT Abi combinations.

We also recognize a serum T and perhaps other serum androgens are prognostic for time to progression and potentially predictive to ARPI's and these just summarizes some of those studies depending upon benchmark ADT by LHRH analog alone does not always achieve castrate levels that we want to see, and especially looking at older studies, 90% will achieve castrate levels at 1.7, but perhaps up to 20 to 30% may not achieve the 0.7. Again, this is dependent upon the type of LHRH analog used and the time of the study and the type of immunoassays, et cetera, but the combinations with ABI usually achieve very low levels of castration and time to CRPC has been linked to serum T levels in a number of studies including PR-7, as well as the ICELAND studies, highlighting the importance of driving T lower and also in the CRPC space, which I won't talk on much at all.

Serum T levels are prognostic for overall survival and may actually identify a subgroup of people who are more likely to respond to abiraterone. So if you have detectable serum androgens, you actually have a better response and longer survival on abiraterone, just illustrating the importance of driving T levels lower.

So why is it an issue? Quantification of T and other serum androgens are challenging at the low levels seen in castrated men. Whereat the limits of quantification of immunoassays, this becomes, again, less of an issue with men on ADT ABI where we're more confident that we are driving T levels lower, below levels of 0.3, but we also recognize after, there other factors beyond quantification that we have to consider ligand bioavailability at the tissue level. Cell level may not always be reflected by serum levels and that relevant ligands differ whether or not you're in the castrate sensitive or in resistance states as genomic mutations occur over time and change AR promiscuity.

So which assay do we use? Historically serum T was measured with an RIA or GC-MS. Now we use, routinely at most institutions, automated immunoassays for total T, which are high throughput, but suited for clinical laboratory environment, but these assays are recognized to be inaccurate at the very low T concentrations due to various interferences that lead to problems with specificity and calibration at low analyte levels.

We have problems quantifying reproducibly at lower limits of quantification with these and so we have to be able to compare this with LC-MS, and I'll move forward a little bit in LC-MS in the future slides, but ELISA's have some pros, they're good sensitivity, they're easy and they fit into a regular laboratory environment, but they do have problems with potential for cross-reactivity and bias, which limits specificity at lower limits and you can only measure one steroid at a time.

LC-MS is the gold standard for specificity and sensitivity and you can also multiplex them, but this is expensive, requires real technical expertise and real specialty expertise within most of laboratories, where at hospitals that are not routinely used in most hospitals.

A number of studies have compared to many T immunoassays against the gold standard and at low levels, immunoassays can give results up to fivefold higher in levels where T is very low and this makes clinical chemists, analytical chemists, very anxious. What's wrong with these assays? How can they even get approved at the FDA? The question of how relevant is this clinically is up to us to determine, but we have to recognize at very low levels you can have poor reproducibility.

Newer assays, for example, ones by Siemens or Roche, have dealt with some of this and they have, at least according to label, precision claims at the levels of castrate detection 0.3 nM variances of about 14 or 15%. So again, quite improved over older assays and it's still important though that these assays still have cross-reactivity other steroids and you can still have two to three fold changes at the very low castrate levels.

Question is, is this relevant? LC-MS/MS is recognized as the most sensitive, specific, precise and accurate measurement of T. It still is subject to errors and again, has to be expertly quantified and implemented.

Overall, imprecision is quite low and the community of analytical chemists who are, one of which is in our university, Dan Holmes, on this recent publication is now collectively trying to harmonize mass specs across the platform and is now able to claim excellent harmonization using standard reference materials with mass specs and these, of course, have the added advantage of being able to multiplex other androgens in one assay.

This is just one example used at one of our hospitals in Vancouver using a Quattro premiere and you can see the levels of sensitivity across a range of different steroids are all very, very low and you can measure this with one assay. Again, it's more expensive than an IA for T, probably about five, six-fold more expensive and not all hospitals have it.

Within the prostate cancer space, a group in Quebec led by Fred Pouliot compared IA's with state of the art LC-MS in 435 samples. They found that the mean T levels were higher with immunoassays than with mass spec. Again, there's a tendency to overestimate with immunoassays. Half the samples with T's greater than 0.7 by IA were actually less than 0.7 by mass spec. So again, clinically, we may think we're not achieving castrate levels, but when you validate with mass spec, you actually are and that's an important interpretive part thing when you look at T's using an IA.

T breakthrough of greater than 0.7, which is deemed to be clinically relevant, was higher with IA than mass spec. So again, mass spec confirmed that T breakthrough is actually less than what you would otherwise think if you are measuring with an IA. In only 3% of samples with a T less than 0.7 by an IA, actually had greater than 0.7 by mass spec. So again, you're unlikely to be missing an elevated T by a low IA measured sample. They recommended suboptimal T levels in castrate patients should be confirmed by either mass spec or validated IA and interpreted with caution before any treatments are made.

A group of Canadian urologists came together and reported in a consensus statement around testosterone suppression, that men receiving ADT for castrate sensitive prostate cancer there is a clinical benefit of achieving a low T level of 0.7 or less and that this is a reasonable goal and men on ADT, this is just LHRH analogs. Prescribers should perform regular T and PSA throughout the first year to ensure adequate T suppression, recognize that IA's may not be sufficiently specific and at least know what method is being used in their institution and know where to get a validated assay if required.

So I'll just conclude by saying that from a why, which and when point of view, lower T levels correlate with better freedom from CRPC, that breakthroughs greater than 1.7 have higher rates of failure and are unacceptable in today's environment. This is, of course, rarely seen as we move combination regimens of abiraterone upstream.

Prescribers should periodically assess T levels to ensure castrate levels perhaps every three to six months, with PSA in the first year, confirm their lab assays accurate at the low levels of T that you're interested in, recognize though that these assays are overestimates of low T and it's more of an issue if they underestimate and this seems to be a rare situation.

I think in the future, we're going to see LC-MS move forward and replace this. It's not ready for prime time yet, but we can anticipate it and certainly should be used for any research studies in this space. The role of T and other serum anogens as prognostic and predictive markers in the CRPC space requires more study as I think there is a potential role for prognostication and prediction there.

Thank you for your attention.