Targeting sRNAs for a precise approach to liquid biopsy

When Rastislav Horos took the job as lab head, now Chief Technology Officer, at Hummingbird Diagnostics in 2020, he faced a tremendous challenge; unstable pre-analytics and underperforming off-the-shelf products. The company had been developing a blood test to detect lung cancer early. But inconclusive data meant that their diagnostic needed to be redesigned.

Three years later, the company’s blood-based screening test to detect lung cancer early, and a complementary diagnostic test for patients with advanced lung cancers, have moved into extensive clinical studies. Both tests rely on measuring the levels of small RNAs (sRNAs), typically 20-40 nucleotides long, including - the well characterized class of microRNAs (miRNAs) that generally regulate gene expression.

Company research has identified altered levels of specific sRNAs in people with lung cancer. Hummingbird chose to focus on sRNAs because their levels fluctuate depending on the tumor's cellular activities. Consequently, small RNA presents a more dynamic and accurate picture of what is happening in the tumors in real-time than other biomarkers that are more static, such as DNA. Moreover, unlike most tests on the market today that only analyze DNA or RNA shed from tumors, Hummingbird’s diagnostic tests also capture the host immune response. 

“We have sRNAs reflecting the immune system, and we have sRNAs clearly from the tumor,” explains Jochen Kohlhaas, Chief Executive Officer of Hummingbird. “We are looking at both, and this is boosting our performance. This is what differentiates us from others.”

How was Hummingbird able to move their research forward? From pre-analytical sample collection to data analysis, company scientists have overhauled their lab protocols to obtain more robust data and are honing their diagnostic tests to make them suitable for clinical settings.

Sample collection

Most companies working in this space rely on blood collected in tubes containing EDTA, a common preservative and anti-coagulant, Horos explains. But most commercially available bio-banked collection tubes don’t stabilize nucleic acids. Additionally, sample handling protocols and storage time till processing can vary among collection sites to such a large extent that it creates bias, Horos adds.

“You can pretty much tell just by the data analysis from which hospital each sample came, and that goes for any technology to analyze nucleic acids, including next-generation sequencing (NGS).”

Hummingbird turned to PreAnalytiX’s PAXgene Blood RNA Tubes for sample collection because they contain reagents that stabilize nucleic acids – they destroy the enzymes that would typically degrade RNA and DNA. They also help keep the pH in the sample low to hinder degradation even further, says Horos. Moreover, Hummingbird optimized the handling and storage protocol of collection tubes.

With the new protocol, “we could completely negate the bias in our NGS data,” says Horos. The feat “was very important to us. We had the pre-analytical issue solved.”

Optimizing discovery by NGS

With extraction of nucleic acid optimized by PAXgene Blood RNA Tubes from the precious samples, the team was ready to move onto small RNA profiling by NGS.

They selected the QIAseq miRNA Library Kit, because it was particularly well suited for biofluids as it eliminates hY4 y RNA, a particularly abundant molecule in this sample source. Empirical data showed the NGS libraries were mostly comprised of just three other miRNAs, says Horos. “These micro RNAs are so highly expressed in red blood cells that they take up 50% of your reads. You can pretty much throw those reads out the window.

Hummingbird teamed up with QIAGEN’s NGS experts in Genomic Services to develop a solution that would enable them to ignore those small RNAs – so-called blocking technology. QIAGEN already offers customers QIAseq FastSelect solution to block other highly abundant RNA molecules like ribosomal RNAs. But Hummingbird needed a customized solution. Based on the company’s recommendations and design, QIAGEN provided custom FastSelect blocking molecules that Hummingbird used in their QIAseq miRNA library preparation.

This solution liberated sequencing bandwidth and enabled a more accurate quantification of less abundant but potentially more informative immune cell-derived miRNAs at a given sequencing depth. Moreover, the company has since expanded the technology to enable the blocking of scores of abundant small RNAs, and further increase the resolution of rarer small RNA species in an NGS experiment.

“By including one single pipetting step and one incubation, we were able to liberate 50% of the reads straight away,” says Horos.

Validation using dPCR

Once Hummingbird optimized sample collection and discovery, they needed to test whether their results with NGS were reproducible with an orthogonal method. They turned to PCR, but they once again faced a technical challenge.

Small RNA genes can encode multiple variants – two may differ by only a single nucleotide, but only one might be informative. Hummingbird tested many different commercially available PCR methods to pluck out specific small RNAs of interest from similar ones – none of the methods worked.

Based on additional research, they developed an in-house PCR assay that had the required sequence specificity. It also reproduced NGS results with high correlation, says Horos. But it had one big drawback: sensitivity. The method was specific but not sensitive – it could not pick up small RNAs with extremely low expression.

Hummingbird again teamed up with QIAGEN to test whether the QIAcuity dPCR system was more sensitive of low-abundance targets. “One of our top RNA biomarkers that we have identified in early cancer signature is not highly expressed. QIAcuity essentially gave us this 1000-fold sensitivity boost that we needed."

Developing diagnostic tests

Hummingbird is applying their technology for the early detection of lung cancer through development of a blood-based screening test, miLung. A recent company study suggested that the miLung signature, comprising both tumor and host response-derived small RNAs, is comparable in performance to low-dose computed tomography – the currently recommended screening method – in identifying people with suspected early lung cancer.

Blocking technology was essential in developing this test given that early stage cancers shed little nucleic acid compared to their more advanced counterparts, and the immune response is mild. The ability to ignore RNAs that are not informative means that signals from tiny amounts of signature RNAs in the sample will be stronger and the test more reliable.

At the same time, a second test is under development to address advanced stage disease at the other end of the lung cancer spectrum, and the question as to what the ideal therapeutic treatment may be. For this complementary diagnostic test, the company has identified a signature of 5 microRNAs in the blood of patients with advanced stage non-small cell lung cancer that can be used to assign patients a risk score, miRisk. 

The goal of miRisk is to more accurately identify advanced stage lung cancer patients who would most benefit from adding chemotherapy to immunotherapy treatment while sparing others the toxic side effects of drugs that would not be effective against their cancer anyway. “Again, the optimized analytical platform, including blocking, was essential to uncover the subtle differences between patients that predicted the ideal therapy combination,” says Horos.

Clinical validation studies for both tests are underway. Hummingbird is transferring their design and protocols for both tests to QIAcuity dPCR machines to enable decentralized testing.  

“Not only does dPCR provide the sensitivity we need, but it also enables de-centralized testing to serve the widest patient communities and advance our fight against the deadliest cancer.”