How advanced liquid biopsy techniques can help determine earlier than ever when cancer immunotherapy is effective – and when it isn’t
George Karlin-Neumann | | Longer Read
At a Glance
- Cancer immunotherapy can be very effective, but many patients do not respond to such treatments
- Immunotherapy’s poor efficacy in these patients is largely due to the imprecision of methods intended to predict who will benefit
- Because of the treatment’s limited success rates and potential severe side effects, oncologists should determine its efficacy as early as possible and adjust treatment accordingly
- Droplet digital PCR technology, which directly quantifies the concentration of circulating tumor DNA, can help
In a field that does not see breakthroughs often enough, immunotherapies have revolutionized cancer care. Instead of the broad, untargeted effects and acquired resistance patients experience with toxic chemotherapy and gene-targeted therapies, immunotherapies can promote a tumor-directed response wherein the patient’s own immune system fights off the cancer. When it works well, patients can achieve deep, long-lasting responses (1).
However, immunotherapies do not work in all patients. This is because the chosen therapy often doesn’t fully address the reason the patient’s cancer has escaped their immune system. The efficacy of checkpoint inhibitors, for example, varies widely across different cancer types, ranging from 15 percent in small-cell lung cancer to 85 percent in Hodgkin’s lymphoma (2,3). Furthermore, immunotherapies can produce severe immune-related adverse effects in patients. And that’s why it’s critical to select the proper therapy for each patient – and to confirm that the patient is seeing a benefit as early as possible after treatment begins.
But most diagnostic tests cannot adequately identify who will benefit from a given immunotherapy. Why? Because current cancer biomarker approaches are not clinically specific and sensitive enough by themselves. Some approaches, such as microsatellite instability testing or estimating tumor mutational burden, involve assessing the likelihood that a tumor is highly antigenic and therefore detectable by the immune system. Others, such as immune gene expression profiling, test whether an inflamed tumor shows evidence of tumor lymphocyte recognition and infiltration. But none of these approaches fully reveals whether or not the patient’s immune system is likely to respond to immunotherapy, nor do they explain why it has not responded on its own.
A well-known example of this is PD-L1 immunohistochemistry tests, which can result in a large number of false positives and false negatives that, in turn, lead to errors in the use of checkpoint inhibitors (4). Not all tumors express high levels of PD-L1, and even the time of sampling and the sample’s location within the tumor may impact the PD-L1 expression levels seen in tests (5). Accurately predicting responders and non-responders prior to treatment – even imperfectly – will almost certainly require combining multiple tests.
Check your answers
Science still has a long way to go before it can accurately predict the optimal treatment regimen for each patient, so oncologists must begin monitoring the patient’s response as early as possible – and, if not favorable, discontinue or adjust treatment accordingly. Traditional diagnostic methods, unfortunately, are unsuitable for this purpose. Regular monitoring requires repeated testing, which makes tissue biopsies risky and impractical – and tissue biopsies may not reveal the genetic characteristics of the entire tumor, nor of secondary metastases (6). This is especially a concern in late-stage cancer, as heterogeneity increases through the course of the disease (7). Blood protein markers are often used to monitor therapy response, but many of these markers, such as serum lactate dehydrogenase for melanoma and prostate specific antigen for prostate cancer, are not very sensitive or specific (8, 9).
Imaging techniques, the current standard of care in assessing response to therapy, provide a phenotypic indicator of cancer progression. They are effective at determining tumor location and size, making cancer staging, tissue biopsies, and surgery possible. But their inability to reveal more than phenotypic information limits their accuracy. For instance, false positives can be caused by pseudoprogression, a phenomenon wherein a tumor appears to be growing when the patient is actually improving. In fact, this “growth” is the result of T cells infiltrating the tumor and causing temporary inflammation. An oncologist could stop or change a patient’s therapy prematurely if they misinterpret this as true progression.
Another common approach to monitoring cancer progression in response to checkpoint inhibitors is to measure protein expression using immunohistochemistry (IHC). PD-1/PD-L1, in particular, is expressed in several tumor types and its presence is often used to guide treatment decisions. But IHC tests for PD-L1 expression are not well-standardized (10); labs may use different antibodies, detection methods, and thresholds, leading to inconsistent results. Furthermore, not all tumors express PD-L1 at the same high levels, and there can even be differences in expression within a single tumor. Thus, although traditional diagnostic techniques reveal useful information about a patient’s tumor, many of them have limitations that make them insufficient or unreliable for monitoring immunotherapy response.
But there is a promising alternative method: liquid biopsy. This approach can often accurately track cancer progression and distinguish who is and isn’t responding to various types of immunotherapy. Using plasma collected from a simple blood draw, liquid biopsies quantify circulating tumor DNA (ctDNA), both a highly specific genetic marker for the tumor and a phenotypic biomarker of successful tumor turnover. Because ctDNA concentration in most cases directly correlates with tumor burden, physicians can distinguish pseudoprogression from true disease progression. Additionally, liquid biopsies are minimally invasive and can be used for serial monitoring with lower risk to the patient. Liquid biopsies can even deliver results within days or weeks following treatment initiation, unlike imaging, which takes place six to 12 weeks after treatment begins.
To monitor cancer progression with liquid biopsies, we must identify tumor-specific ctDNA mutations. If the tumor’s mutations are unknown, next-generation sequencing (NGS) is a comprehensive method for initially profiling tumor-associated genetic mutations in the blood. Once the mutations are determined, tests based on droplet digital PCR (ddPCR) and related technologies can quickly and cost-effectively quantify and track these mutations in blood or other body fluids as biomarkers of treatment response and disease progression.
ddPCR liquid biopsy in practice
Several investigations over the past few years have demonstrated the ability of ddPCR-based tests to distinguish responders from non-responders, often within weeks of treatment initiation. This capability has been demonstrated for several different types of immunotherapies across multiple cancer types, including melanoma, non-small cell lung cancer (NSCLC), and cervical cancer. Specifically, several studies have shown how a liquid biopsy based on ddPCR can predict and monitor the effectiveness of checkpoint inhibitors – currently the most widely used class of immunotherapies – as well as CAR T cells and tumor infiltrating lymphocytes (TILs), which are expected to be a major part of the next generation of immunotherapies.
In one case, Jenny Lee and colleagues developed a ddPCR liquid biopsy to track BRAF,NRAS,and KIT mutations. They evaluated the test’s ability to distinguish between responders and non-responders to anti-PD-L1 (+/- anti-CTLA-4) therapy in 105 patients with stage IV melanoma. They assayed these mutations in both training (n=76) and validation (n=29) cohorts at the start of therapy and at regular intervals for up to 12 weeks and found that longitudinal monitoring of ctDNA was an effective means to identify patients who responded to the checkpoint inhibitors (11). Ultimately, in this study, most patients who did not have detectable ctDNA by 12 weeks survived to at least a median of 17.5 months following the start of therapy. In contrast, patients who still had detectable ctDNA after 12 weeks had poorer outcomes, with a median overall survival of 9.7 months.
The researchers followed this with another study of ctDNA monitoring, this time to evaluate its ability to identify pseudoprogression in a 125-patient cohort, 29 of whom were identified by CT scans as having progressive disease (12). In contrast to the imaging results, ctDNA profiles measured using ddPCR-based liquid biopsy revealed that nine of these 29 patients had favorable ctDNA profiles and were actually exhibiting pseudoprogression. They also correctly identified unfavorable ctDNA profiles in 18 of the remaining 20 patients. Consequently, ctDNA monitoring could accurately separate those who exhibit pseudoprogression from true progressors who might benefit from changing or discontinuing treatment.
Researchers at the Groningen University Medical Center in the Netherlands recently reported the use of a similar technique to measure ctDNA levels of KRAS exon 2 mutations in 16 (since expanded to 29) NSCLC patients undergoing nivolumab (anti-PD-1) treatment (13). Patients with a positive response to nivolumab, an anti-PD-1 treatment, showed a distinctive ctDNA kinetic profile in which ctDNA levels spiked one week after the start of therapy and dropped to undetectable levels a week later. Additional measurements in the first three to seven weeks validated the initial findings. Conversely, patients whose tumors did not respond to nivolumab (as evidenced by increasing RECIST 1.1 scores) showed steadily increasing levels of ctDNA in their blood.
Another biomarker under evaluation for immunotherapy response is PD-1 mRNA in exosomes released by cancer cells into the blood. Marzia Del Re and colleagues at the University of Pisa demonstrated that a liquid biopsy based on ddPCR could track melanoma and NSCLC progression in patients treated with the PD-1 inhibitors nivolumab and pembrolizumab after only two months of treatment (14). Among the 26 patients in the study, PD-L1 mRNA levels decreased by an average of 71 percent in patients who entered complete or partial remission, whereas levels increased by an average of 104 percent in patients with progressive disease.
Beyond checkpoint inhibitors, ddPCR liquid biopsy has also been used to monitor responses to adoptive cell therapy approaches such as CAR T cell therapy and tumor-infiltrating lymphocyte (TIL) therapy. In CAR T therapy, ddPCR liquid biopsy could potentially monitor the persistence of CAR T cells, which, in preclinical models, is a predictor of overall survival in cases of acute myeloid leukemia (AML). CAR T therapy involves removing a patient’s T cells, genetically modifying them to express a chimeric antigen receptor (CAR), and injecting them back into the patient to attack tumor cells. For CAR T cells to work in the long term, however, they must persist in the body for months in an intermediate concentration; too high a dose can be toxic, but too low a dose may not be effective.
Mayumi Sugita of Weill Cornell Medicine found that ddPCR liquid biopsy very effectively monitored the persistence of CAR T cells (15). Typically, CAR T cell kinetics are followed using multi-parameter flow cytometry (MFC), but this technique is hard to validate because it is less sensitive than liquid biopsy. Sugita and her team were able to predict overall survival in a cohort of 20 patient-derived xenograft mice with established human AML by simultaneously monitoring minimal residual disease (via the NPM1 tumor mutation) and CAR T cell persistence using ddPCR liquid biopsy. In fact, the ddPCR method was able to detect CAR T cells that could not be evaluated by MFC in peripheral blood.
TILs are another type of immunotherapy that has long been studied as a potential option for treating several different cancers, including cervical cancer (16). In a study examining the validity of using human papilloma virus (HPV) cell-free DNA (cfDNA) to monitor cervical cancer, Zhigang Kang from the National Cancer Institute found that cfDNA exhibits a distinct pharmacokinetic response to TIL therapy; in a cohort of nine patients, the three that experienced objective regression exhibited a spike in HPV cfDNA after two or three days, followed by a drop to undetectable levels in about a week (17). This kinetic profile reflected the anti-tumor activities of the TILs, suggesting that ddPCR-based testing can be used to monitor their effectiveness within a few days of TIL administration.
The future of ddPCR-based liquid biopsies
Liquid biopsies provide sensitivity, accuracy, and reliability where current pre-treatment predictive tests may not – but they still have their limitations. Because ddPCR relies on circulating genetic material to monitor cancer progression, it may be less reliable for monitoring tumors in locations where they cannot slough ctDNA into the blood (for instance, intracranial tumors). In most cases, though, ctDNA is a more direct indicator of tumor load than imaging or immunoassays – and, because liquid biopsies are minimally invasive, they can generally serve as a dependable method for serial monitoring of immunotherapy effectiveness. Ultimately, this method could enable physicians to check their initial treatment decisions early and often, and to adjust or change each patient’s therapy to maximize their chances of living a long and healthy life.
- NCI Staff, “Checking in on cancer checkpoint inhibitors” (2015). Available at: bit.ly/2lfjvU0. Accessed April 18, 2019.
- SJ Antonia et al., “Phase I/II study of nivolumab with or without ipilimumab for treatment of recurrent small cell lung cancer (SCLC): CA209-032”, J Immunother Cancer, 3, 376 (2015).
- ER Plimack et al., “Pembrolizumab (MK-3475) for advanced urothelial cancer: Updated results and biomarker analysis from KEYNOTE-012”, J Clin Oncol, 33, 4502 (2017).
- M Udall et al., “PD-L1 diagnostic tests: a systematic literature review of scoring algorithms and test-validation metrics”, Diagn Pathol, 13, 12 (2018). PMID: 29426340.
- D Escors et al., “The intracellular signalosome of PD-L1 in cancer cells”, Signal Transduct Target Ther, 3, 26 (2018). PMID: 30275987.
- C Mao et al., “Blood as a substitute for tumor tissue in detecting EGFR mutations for guiding EGFR TKIs treatment of nonsmall cell lung cancer: a systematic review and meta-analysis”, Medicine, 94, e775 (2015). PMID: 26020382.
- I Dagogo-Jack, AT Shaw, “Tumour heterogeneity and resistance to cancer therapies”, Nat Rev Clin Oncol, 15, 81 (2018). PMID: 29115304.
- SC Tsao et al., “Monitoring response to therapy in melanoma by quantifying circulating tumour DNA with droplet digital PCR for BRAF and NRAS mutations”, Sci Rep, 5, 11198 (2015). PMID: 26095797.
- M Adhyam, AK Gupta, “A review on the clinical utility of PSA in cancer prostate”, Indian J Surg Oncol, 3, 120 (2012). PMID: 23730101.
- L Li et al., “Promising clinical application of ctDNA in evaluating immunotherapy efficacy”, Am J Cancer Res, 8, 1947 (2018). PMID: 30416847.
- JH Lee et al., “Circulating tumour DNA predicts response to anti-PD1 antibodies in metastatic melanoma”, Ann Oncol, 28, 1130 (2017). PMID: 28327969.
- JH Lee et al., “Association between circulating tumor DNA and pseudoprogression in patients with metastatic melanoma treated with anti–programmed cell death 1 antibodies”, JAMA Oncol, 4, 717 (2018). PMID: 29423503.
- TJN Hiltermann et al., “ctDNA a promising predictive marker for treatment with PD-1 inhibitors in KRAS mutated NSCLC after platinum based chemotherapy”. Poster presented at the AACR Annual Meeting; March 31, 2019; Atlanta, USA. Poster #399.
- M Del Re et al., “PD-L1 mRNA expression in plasma-derived exosomes is associated with response to anti-PD-1 antibodies in melanoma and NSCLC”, Br J Cancer, 118, 820 (2018). PMID: 29509748.
- M Sugita et al., “Prediction of immunotherapy outcome by multimodal assessment of minimal residual disease and persistence of allogeneic anti-CD123 CAR T-cells (UCART123) in pre-clinical models of acute myeloid leukemia”. Poster presented at the AACR Annual Meeting 2018; April 14–18, 2018; Chicago, USA. Abstract #5681.
- BC Sheu et al., “Predominant Th2/Tc2 polarity of tumor-infiltrating lymphocytes in human cervical cancer”, J Immunol, 167, 2972 (2001). PMID: 11509647.
- Z Kang et al., “Circulating cell-free DNA for metastatic cervical cancer detection, genotyping, and monitoring”, Clin Cancer Res, 23, 6856 (2017). PMID: 28899967.