Issue 04 | April 2017

Welcome to Strand Genomics-A Monthly E-zine from Strand Life Sciences

Strand Life Sciences welcomes you to Strand Genomics, our monthly E-zine that includes articles of interest to physicians. This e-zine brings the latest news in the world of genetic diagnostics, to your doorstep. The E-zine features carefully crafted articles as well as curated news in the field of cancer therapy and genetic analyses to support the implementation of personalized medical care. We invite you to peruse as well as share these articles. Please also feel free to write back to us with comments and questions at strandlive@strandls.com

Circulating Tumor DNA: Better Marker Than Histological Grading of Tumors?

Dr. Shefali Sabharanjak
Strand Life Sciences

Abstract

  • Liquid biopsy enabled analysis of cfDNA may offer a chance to add a new criterion in the initial diagnosis of cancer
  • The presence of high cfDNA or ctDNA can be potentially used as a measure of the pathogenicity of cancer in addition to other parameters like cellular morphology, degree of dedifferentiation and metastasis. Several publications (Garcia-Murillas et al. 2015; Tie et al. 2016; Ai et al. 2016; Olsson et al. 2015) have alluded to the fact that the cfDNA yield and ctDNA levels are indicative of advanced and metastatic disease, though it has not been specifically established in a study.
  • In a tumor – plasma concordance study conducted by Strand Life Sciences, we have identified cases of cancer wherein the ctDNA load was 10-100-fold higher than the most commonly observed rate of incidence.
  • Interestingly, the histological data on several of these cases classified them as Stage I and II and clearly not advanced disease, and treatment was conducted accordingly. In the cases where follow-up data was available, the patients were lost to disease, soon after diagnosis.
  • Evidence from other longitudinal follow-up studies also indicates that identification of ctDNA from cancer patients may have better prognostic value than other clinical and pathological indicators.
  • The data argues for inclusion of liquid biopsy as a monitoring tool in the initial diagnosis as well as treatment of cancer patients.

Introduction

Cell-free DNA (cfDNA) shed by dying cells from normal tissues as well as tumor DNA (also referred to as circulating tumor DNA or ctDNA) from cancer cells that can be isolated and analyzed from the plasma fraction of blood is a goldmine of information about the status of cancer in a patient. In fact, data from some recent studies suggests that assessment of ctDNA during the course of cancer therapy is a better indicator of the progression of disease than other biomarkers (Garcia-Murillas et al. 2015; Ai et al. 2016).

Strand Life Sciences has developed liquid biopsy tests for the detection of mutations in cfDNA from plasma samples of cancer patients. We have carried out extensive validation of our ultrasensitive liquid biopsy tests using solid tumor tissue as well as a liquid biopsy sample to assess concordance between results. In addition to establishing the validity of our lab-developed liquid biopsy tests, the concordance study yielded some surprising observations on patient prognosis.

Elevated ctDNA in samples

As a general observation, we are able to detect 10-40 copies of ctDNA from 33-40 ng of cfDNA (which corresponds to 10,000 genome equivalents as per parameters established in our lab). There have been however unusual cases wherein the amount of ctDNA detected per 10,000 genome equivalents has far exceeded this normal range (Table 1).

Table 1. Elevated levels of ctDNA in patients with early tumors as graded by histopathology

 Patient ID (Pseudonym)  Type of Cancer  No. of copies of ctDNA / 10,000 genome equivalents Histopathological Assessment  Patient status
 StranBladCa-02 Bladder Cancer 736 Poorly differentiated carcinoma favoring invasive urothelial (transitional) carcinoma pT1pNxpMx  Expired within 12 months after diagnosis of cancer
 StranKidneyCA-05 Bladder Cancer 4120 Invasive transitional urothelial cell carcinoma with squamous differentiation, grade3; pT4,pN2 Expired shortly after diagnosis of cancer
 StranCH-FL1  5360  Expired after one round of radiotherapy
 StranLungCa-01  Lung cancer  3120  Expired
 StranCRC-02  Colorectal cancer  56 at T0, 42 at follow-up  Expired
StranOvCa-01 Ovarian cancer 136  Expired

Patient StranBladCa-02 presented with bladder cancer that was classified as an early stage tumor based on standard histopathology. However, the amount of ctDNA isolated from the patient was surprisingly high: 736 copies of ctDNA / 10,000 genome equivalents. Studies have shown that the amount of ctDNA shed by tumors usually correlates with the stage of the tumor (Bettegowda et al. 2014). Early stage tumors shed lower amounts of ctDNA whereas the amount of ctDNA shed by metastatic cancers is higher. Surprisingly, in this case analyzed at Strand, although the histopathological classification of the tumor was “non-invasive” (pT1pNxpMx), the amount of ctDNA found in plasma fractions was significantly higher.
Upon follow up at one year post surgery, it was found that the patient expired soon after the diagnosis of the early stage cancer.

Likewise, in another kidney cancer case, StranKidneyCa-05, the ctDNA load in the plasma was extraordinarily high: 4120 copies per 10,000 genome equivalents. In this case, the histopathological characterization revealed that the tumor had infiltrated the kidney as well as perinephric adipose tissue. However it was classified as grade3; pT4, pN2 and there was no indication of metastasis. ctDNA in the plasma sample from this patient was nearly 100-fold higher than the average values. The patient was lost to the disease shortly after diagnosis.

A follow-up of other cases with unusually high amounts of ctDNA also revealed that the patients succumbed to cancer within a short period following the initial diagnosis of their cancers.

Although these are early days yet, the data seems to indicate that ctDNA may perhaps be a better prognostic marker than histopathological indicators of tumor differentiation. Alternatively, high amounts of ctDNA could indicate the presence of cancer cells at cryptic locations, other than the tissue of origin. Sole reliance on histological classification of an early stage tumor may create a false diagnosis of a tumor that is perhaps amenable to therapeutic options that are deemed as standard of care.

The predictive ability of ctDNA has been assessed in other studies as well(Tie et al. 2016). In this study, 178 patients with stage II colorectal cancer were profiled for tumor markers using liquid biopsy. The presence of ctDNA 4-10 weeks after surgery was strongly indicative of relapse in the 2 yr time period (Figure 1).

Figure 1. Stratification of colorectal cancer patients by ctDNA vs clinicopathological risk

all no chemo patients

no chemo patients
(Ref: (Tie et al. 2016))

Fourteen patients showed presence of ctDNA in the plasma in this cohort. Eleven patients from this group ( 78.4%) showed recurrence of the cancer by radiological detection methods. As can be seen from Figure 1, the incidence of ctDNA in the plasma is a better prognostic factor than other clinicopathological factors.

Summary

  • Liquid biopsy offers a chance to add a new criterion in the diagnosis of cancer.
  • The presence or absence of ctDNA can be potentially used as a measure of the pathogenicity of cancer in addition to other parameters like cellular morphology, degree of dedifferentiation and metastasis.
  • At Strand Life Sciences, we have identified some cases of cancer wherein the ctDNA load was 10-100-fold higher than the most commonly observed rate of incidence.
  • Although the tumors from these patients were histologically classified as early stage tumors, the patients were lost to disease, soon after diagnosis.
  • Evidence from other longitudinal follow-up studies also indicates that identification of ctDNA from cancer patients may have better prognostic value than other clinical and pathological indicators.
  • The data argues for inclusion of liquid biopsy as a monitoring tool in the initial diagnosis as well as treatment of cancer patients.

References

Ai, B. et al., 2016. Circulating cell-free DNA as a prognostic and predictive biomarker in non-small cell lung cancer. Oncotarget, 7(28), pp.44583–44595. Available at: http://www.ncbi.nlm.nih.gov/pubmed/27323821 [Accessed February 16, 2017].

Bettegowda, C. et al., 2014. Detection of circulating tumor DNA in early- and late-stage human malignancies. Science translational medicine, 6(224), p.224ra24. Available at: http://www.ncbi.nlm.nih.gov/pubmed/24553385 [Accessed January 7, 2017].

Garcia-Murillas, I. et al., 2015. Mutation tracking in circulating tumor DNA predicts relapse in early breast cancer. Science Translational Medicine, 7(302), p.302ra133-302ra133. Available at: http://www.ncbi.nlm.nih.gov/pubmed/26311728 [Accessed March 13, 2017].

Olsson, E. et al., 2015. Serial monitoring of circulating tumor DNA in patients with primary breast cancer for detection of occult metastatic disease. EMBO Molecular Medicine, 7(8).

Tie, J. et al., 2016. Circulating tumor DNA analysis detects minimal residual disease and predicts recurrence in patients with stage II colon cancer. Science Translational Medicine, 8(346), p.346ra92-346ra92. Available at: http://www.ncbi.nlm.nih.gov/pubmed/27384348 [Accessed February 16, 2017].

Show full article.

Strand Liquid Biopsy Tests

Strand Life Sciences has recently launched a portfolio of liquid biopsy tests for detection of mutations associated with cancer from circulating tumor DNA. Strand is known for introducing innovations in the healthcare space that are backed by research conducted and validated by Strand’s scientists. In this video, Dr. Ramesh Hariharan, CEO- Strand Life Sciences elaborates on the analytical validation of Strand’s liquid biopsy tests. Watch now!

Strand Liquid Biopsy test video

NGS-based Genetic Analyses: All Tests Are Not Equally Efficient

Dr. Shefali Sabharanjak
Strand Life Sciences

Abstract

  • Results of genetic testing obtained from various laboratories may differ in quality and therefore in identification of gene variants.
  • Strand Life sciences provides genetic tests that are ably supported by custom-designed bioinformatics software pipeline.
  • High coverage of the tested genomes, identification of multiple types of genomic alterations and quality-testing at 26 checkpoints are some of the features of genetic testing conducted at Strand.
  • Strand’s genomic analyses workflow has been periodically audited and validated as per requirements of the College of American Pathologists (CAP).
  • Inexpensive genomic tests are likely to yield inaccurate results due to exclusion of controls that otherwise add to cost.
  • Strand has protocols built-in to identify and eliminate several possible errors in NGS-based genetic analyses.

Introduction

Genetic analyses are quickly taking center stage in the diagnosis as well as management of several diseases. Companion tests for management of cancer are amongst the most prominent applications of genetic testing. However, genetics-aided diagnostics and therapeutic applications are emerging in other areas such as cardiovascular medicine and neurology as well (Caraco et al. 2008; Malhotra et al. 2016; Otto et al. 2017).

Technological advances enabling the rapid sequencing of DNA have greatly facilitated the penetration of genetic testing facilities in healthcare settings. We envisage that within the next 10 years, more than 50% of therapeutic decisions are aided by appropriate genetic tests.

A proliferation in providers of genetic analyses is expected in order to keep up with the demand for genetic tests.
There is a plethora of software tools that are designed to solve analytical challenges that are thrown up by large-scale sequencing technologies like Next-Generation Sequencing (NGS).

So, as a physician, how do you decide between an accurate NGS test and an inaccurate one? There are some critical components of a good genetic result obtained by NGS techniques, that can help in making these distinctions.

Coverage of the Genes of Interest

One prominent feature of a good NGS test is the extent of coverage of the genes of interest, stated in the test report. Coverage essentially refers to the number of times a particular stretch of DNA has been ‘read’ or sequenced. This is important in order to ascertain that the readout of a base (or bases) that is obtained in that stretch of DNA is statistically accurate. At coverage level less than 20X, the chances of obtaining a false positive, due to sequencing errors, are high. Likewise, the chances of obtaining a false negative, because of variant calls not being sufficiently covered are also higher at lower coverage of the genes. Costs of sequencing would be higher to obtain higher coverages. So, although a low coverage test is cheaper, the chances of an erroneous result are higher with such tests. (The coverage requirements would be 10 times more for accurately calling out variants for somatic cases to achieve high level of confidence in calling out the variants.)

Figure1. Coverage of a stretch of DNA in NGS analysis


Scientists believe that the least acceptable coverage for an accurate diagnosis is 20X, a standard adopted at Strand, for all germline tests. In order to obtain 20X coverage with all the DNA probes used in a particular genetic test, the overall coverage (averaged across all the gene regions sequenced) of the test needs to be more than 200 X. The number of bases that can be potentially missed out is minimized at this extent of coverage. These coverage requirements have been established during our extensive validation studies conducted for every panel (which is essentially a set of DNA probes) individually using control samples that have clinically relevant variants.

So, what can happen if we reduce coverage? Even if the extent of coverage is reduced by half, the chances of missing out on correct identification of a base pair are doubled. Hence, while NGS tests with lower coverage will definitely be cheaper and will provide a fast result, the accuracy of such tests is going to be only 50%. At Strand Life Sciences, all our tests are designed and validated to operate at 200-500X coverage- eliminating the possibility of obtaining both false positive as well as false negative readouts!

Validation of NGS Readouts by Sanger Sequencing

Why are we so sure that that a coverage level of 200-500X works for Strand’s NGS panels? For starters, we have validated the NGS readouts of 295 mutations across 102 genes by Sanger sequencing. The results from both techniques are 100% concordant, indicating that our NGS test coverage is not only adequate but also yields accurate results.

Figure 2. Validation of NGS Readouts by Sanger Sequencing

NGS sanger sequencing
So, what about the remaining 4% genes missed out? These were due to the variants not satisfying our QC acceptance criteria or at the borderline of acceptable QC values. For all such variants, we validate the variant using Sanger sequencing or any other appropriate method, as a good follow-up measure. Notifying the physician of a poor quality of a variant and confirmation of the same, wherever possible, are good lab practices we follow to give more confidence to physicians to take actions based on our reports.

Large Genomic Insertions And Deletions

All genomic changes need not be only single nucleotide variants. Some genomic alterations like large insertions or deletions of DNA may also be present in the sample provided. Sometimes, genes like BRCA1 and RB1 may have entire exons missing (deletions) or have more than the normally expected copies (duplications)(Singh et al. 2016). This is where the StrandNGS software is distinct from other out-of-the-box analytical software. Strand’s bioinformatics software is designed to identify deletions as well as duplications in stretches of DNA. Additionally, we offer confirmatory tests like MLPA (Multiplex ligation-dependent Probe Amplification) detection for certain genes like BRCA1/2 and RB1, voluntarily, in case the quality metrices for these copy number calls (deletion or duplication) do not satisfy our acceptability criteria. So far, we have validated 94 cases, involving 6 genes, using MLPA and have had no false positives.

We also confirm negative results for indels by Sanger sequencing, voluntarily.

These values are a pretty good indication of the high accuracy of Strand’s NGS tests.

Navigating Around Other Potential Errors

In addition to accurate detection of single nucleotide variants and large insertions and deletions, there are other things that could go wrong with genomic tests. Sample swaps, repetitive analysis of the same sample, contamination and unpredictable errors are some more hurdles that can compromise the results of the test. Strand scientists have established an array of 26 such checkpoints that are applied to every run of NGS sequencing in order to minimize such human as well as incidental errors.

Validation of Strand’s Genomic Analyses by CAP

We also undergo periodic proficiency testing at the hands of College of American Pathologists (CAP), an organization that examines quality metrics implemented at diagnostic laboratories, worldwide. Essentially, CAP sends de-identified (blinded), identical clinical samples to labs such as Strand, worldwide and then compares the results of their tests. Strand has scored 100% in all the proficiency tests administered by CAP. For the record, the performance of other labs, from all over the world, ranges between 85-100%.

Additionally, CAP has also audited our laboratory workflow and the fact that we run clinical samples with appropriate controls using the same parameters that have been established during our validation runs.

Summary

In conclusion, a good NGS test (germline as well as somatic DNA analyses) should have:

  • Appropriately High Coverage
  • Confirmation of variants with poor QC metrices by Sanger Sequencing or an alternative second method.
  • Multiple checkpoints in wet lab as well as bioinformatics workflow
  • Ability to identify simple (Single Nucleotide Variants) as well as complex (indels and duplications) genomic changes
  • Validation to establish assay performance characteristics and continuous monitoring of the assay performance by use of appropriate controls and participation in proficiency testing by national and internationally reputed accreditation organization like NABL and CAP.

NGS-based Genetic Analyses summary

References

Caraco, Y., Blotnick, S. & Muszkat, M., 2008. CYP2C9 Genotype-guided Warfarin Prescribing Enhances the Efficacy and Safety of Anticoagulation: A Prospective Randomized Controlled Study. Clinical Pharmacology & Therapeutics, 83(3), pp.460–470. Available at: http://www.ncbi.nlm.nih.gov/pubmed/17851566 [Accessed April 6, 2017].

Malhotra, S. et al., 2016. Candidate genes for alcohol dependence: A genetic association study from India. Indian Journal of Medical Research, 144(5), p.689. Available at: http://www.ncbi.nlm.nih.gov/pubmed/28361821 [Accessed April 6, 2017].

Otto, J.M. et al., 2017. Genetic Variation in the Exome: Associations With Alcohol and Tobacco Co-Use. Psychology of Addictive Behaviors. Available at: http://www.ncbi.nlm.nih.gov/pubmed/28368157 [Accessed April 6, 2017].

Singh, J. et al., 2016. Next-generation sequencing-based method shows increased mutation detection sensitivity in an Indian retinoblastoma cohort. Molecular vision, 22, pp.1036–47. Available at: http://www.ncbi.nlm.nih.gov/pubmed/27582626 [Accessed April 10, 2017].

Show full article.