LIWUID BIOPSIES READY FOR PRIME TIME?

The key comparisons were for the sensitivity of plasma and urine testing for T790M vs tissue testing as a reference. Among 387 patients positive for T790M on tissue testing from the 482 for whom both plasma and tissue samples were available for correlative testing, the sensitivity of plasma testing was 80.9% (313 of 387). When plasma testing results in patients who had tissue that tested negative or had an inadequate biopsy sample (a common event in clinical practice) were looked at, plasma testing was able to identify 374 patients—minimally less than the 387 identified by tissue.

Results with urine were remarkably similar, although a much smaller subset of only 213 patients had paired samples for both plasma and urine testing. Among them, the sensitivity of urine was 81.1% (142 of 175 with positive findings on tissue testing). However, when the larger number of patients in whom the tissue sample was negative or inadequate for testing were considered, urine identified 169 patients as T790M-positive—a nearly identical number to that of tissue testing.

When the range of results in 181 patients who had testing of tissue, plasma, and urine samples was examined, each modality identified a nearly identical number but incompletely overlapping population of patients as T790M-negative (146 by tissue, 145 by plasma, and 144 by urine testing). This means that although each had a sensitivity of approximately 80%, these testing modalities were complementary, rather than redundant. The best way to identify patients as T790M-positive is to combine two or even three different methods.

A subset analysis looked at whether the sensitivity of liquid biopsy testing techniques correlated with the location and extent of disease, reporting that in patients with M1a disease, representing lung cancer limited to the chest, sensitivity was somewhat lower when both plasma (56.8%; P < .001) and urine (73.8%; P = .12) samples were used. This suggests that "tumor burden" and perhaps more distant spread can correlate with the detectability of circulating tumor DNA.

The confirmed response rate to rociletinib in the overall population was 33.9%, but a critical question is whether patients who were identified as positive by the different testing modalities were equally likely to respond. In fact, response rates were similar, at 32%-37%, regardless of whether T790M was detected by tissue, plasma, or urine. Similarly, duration of response and progression-free survival were very similar regardless of how T790M was detected.

Although development of rociletinib is being discontinued, taken together the mounting data supporting the use of plasma and even urine to detect T790M makes either of these "liquid biopsy" options viable alternatives or supplements to a repeat tissue biopsy for EGFR mutation-positive patients with acquired resistance, for whom osimertinib is now an approved and highly active therapy for the significant subset of patients with progressing cancer that harbors the T790M acquired resistance mutation. This approach is uniquely appealing for patients with EGFR mutation-positive lung cancer, who often have relatively inaccessible disease, and where you are looking for a very distinct, clinically relevant target in T790M.

Coupled with the very recent approval of a plasma test for initial detection of an activating EGFR mutation for first-line treatment with erlotinib,^[8] these data may arguably bring us to a tipping point in which liquid biopsy testing becomes widely used in these contexts. In addition, other testing approaches even offer the potential for next-generation sequencing to look for a broad panel of potentially actionable mutations in patients for whom tissue testing is not readily available—a setting that awaits further study before becoming a standard approach.

For now, converging evidence and the growing availability of liquid biopsy techniques make plasma and now even urine testing a viable alternative to repeat biopsies—a development that we can expect will be eagerly welcomed by physicians and patients alike.

As we consider data from the American Society of Clinical Oncology 2016 annual meeting that may be practice-changing, one of the most immediately clinically relevant results may have been work that highlighted the potential utility of "liquid biopsy" techniques in the setting of acquired resistance for patients with an activating mutation in the epidermal growth factor receptor (EGFR) gene.

Heather Wakelee, MD, from Stanford University Medical Center, reported molecular correlate work comparing testing strategies using tissue and liquid samples, including both plasma and urine, looking for the acquired resistance mutation T790M.^[1]This mutation is seen in approximately 50%-60% of patients with EGFR mutation-positive acquired resistance^[2] and is associated with a high probability of responding to the third-generation EGFR tyrosine kinase inhibitors osimertinib,^[3] which is now approved by the US Food and Drug Administration,^[4] and rociletinib,^[5] which is no longer being studied for further development.^[6]

Seen in the context of the previously reported TIGER-X trial of rociletinib in EGFR mutation-positive patients, this particular work focused on molecular testing techniques.^[5,7] Although clinical results with the ill-fated rociletinib are of only marginal interest now that this agent is consigned to a historic footnote, its commercial death may not have been in vain: Results from this presentation were a highlight in lung cancer that offered compelling evidence that plasma and even urine testing for T790M has a sensitivity that rivals that of tissue testing. Paired with the overwhelmingly greater accessibility of plasma and especially urine samples, such "liquid biopsy" techniques have become increasingly available commercially but have awaited sufficient clinical data to validate their routine use.

The patient population comprised 548 EGFR mutation-positive patients with safety data, including a subset of 443 with data on the efficacy of rociletinib. Among these patients, 540 had tissue testing, 482 had plasma available for correlation, and 213 had urine available as part of a later protocol amendment. Plasma testing was conducted using a quantitative and sensitive BEAMing test (Sysmex Inostics; Mundelein, Illinois), consisting of digital polymerase chain reaction followed by flow cytometry^[7] that was capable of identifying a limited array of EGFRmutations, including T790M, with very high sensitivity. Urine testing used quantitative next-generation sequencing assay (Trovagene; San Diego, California), which is also able to detect T790M along with the activating deletion 19 and L858R substitution mutations with very high sensitivity