Feasibility of liquid biopsy using plasma and platelets for detection of anaplastic lymphoma kinase rearrangements in non-small cell lung cancer

FISH assays for the detection of ALK rearrangement were performed using tumor specimens from 1,128 patients with NSCLC between January 2015 and June 2018. We retrospectively analyzed formalin-fixed paraffin-embedded (FFPE) tissues from previously confirmed FISH-positive (n = 199) and -negative (n = 920) cases. We recruited patients who had available tissue specimens and agreed to venous sampling (Fig. 1). Tumor samples were obtained from FFPE sections from tumor tissues and FFPE cell blocks from cytology specimens obtained from bronchoscopic procedures (washing or brushing) or malignant pleural effusions. Tumor samples for RT-PCR were collected from the same specimens used for FISH assays. Blood samples were obtained at diagnosis, before or during systemic treatment with cytotoxic chemotherapy and ALK TKIs. For analysis of serial monitoring during ALK TKI treatment, enrolled patients administered ALK TKIs were required to provide a blood sample at every visit, and venous sampling was performed on weeks 1, 4, and 8 and then every 2–3 months in parallel with imaging studies for response assessment.

ALK rearrangement was detected by FISH assays using a break-apart probe specific for the ALK locus (Vysis LSI ALK dual-color, break-apart rearrangement probe; Abbott Molecular, Abbott Park, IL, USA) in FFPE tumor tissue samples. FISH-positive samples were defined as those with more than 15% of tumor cells showing split signals or an isolated red signal (3′ signal) as described in the previous studies (Fig. 2a) (Paik et al. 2011; Shaw et al. 2009).

Plasma and platelets were isolated from the same sample of whole blood in 10-mL purple-cap BD Vacutainers containing ethylenediaminetetraacetic acid (EDTA) anticoagulant, and Eppendorf centrifugation (5810R) was carried out immediately before storage within 2 h from venous sampling. The cells and aggregates were removed by centrifugation at 4 °C for 10 min at 700×g, yielding platelet-rich plasma. The platelets were isolated from the platelet-rich plasma by centrifugation at 4 °C for 10 min at 1700×g, and the plasma was then collected as an aliquot from the supernatant and frozen at − 80 °C for further use. The platelet pellet was collected and mixed with 300 μL RNAlater solution (Life Technologies, Carlsbad, CA, USA) and frozen at − 80 °C for further use. The plasma and platelets were frozen in parallel.

In case of FFPE tissues, RNA was extracted from 3 slides of 4–5–μm thickness taken from the FFPE blocks using the PureLink™ FFPE Total RNA Isolation Kit (Invitrogen, Carlsbad, CA, USA). RNA was also extracted from the plasma and platelets using a RiboPure-Blood Kit (Life Technologies) according to the manufacturers’ instructions. The resulting RNA was eluted in 50-μL elution buffer. The concentration and purity of the extracted RNA were determined using a NanoDrop ND-2000 spectrophotometer (Thermo Fisher Scientific, DE, USA). The quantity of total extracted RNA was only measured in FISH-positive patients. The extracted RNA was stocked at − 80 °C until use. We used 250-ng total RNA for cDNA synthesis using a SuperScript VILO cDNA Synthesis Kit (Life Technologies).

The most common ALK rearrangement is characterized by fusion of the ALK gene with echinoderm microtubule-associated protein-like 4 (EML4); the fusion gene has multiple chimeric variants with the same portion of ALK and different truncations of EML4 (Sasaki et al. 2010; Soda et al. 2007). The EML4-ALK fusion RNA was detected by PNA-mediated RT-PCR assays using a PANA EML4-ALK Fusion Gene Detection Screening Kit (Panagene, Daejeon, South Korea), which was initially developed for tissue genotyping. This assay was previously performed in a retrospective study for genotyping of ALK fusion variants and showed favorable performance in terms of sensitivity (57/81, 70.4%) and meaningful correlations with clinical implications, demonstrating that specific variants of EML4-ALK showed poor prognosis and multidrug resistance to ALK TKI treatment (Woo et al. 2017). This kit was designed to detect 28 types of known ALK rearrangements for screening, including E6;A19, E6;A20 (variant 3a), E6ins33;A20 (variant 3b, three subtypes), E6;ins18A20, E13;A20 (variant 1, five subtypes), E13;ins69A20 (variant 6, two subtypes), E20;A20 (variant 2, two subtypes), E20; ins18A20 (two subtypes), E14ins11; del49A20 (variant 4), E14; del12A20 (variant 7), E14; del36A20, E14ins2; ins56A20, E2; A20 (variant 5a), E2; ins117A20 (variant 5b), E17; ins30A20 (variant 8a), E17ins61; ins34A20 (variant 8b), E17ins65; A20, E17; ins68A20, and E17del58; ins39A20. PCR was performed under the following conditions: 2 min at 50 °C; 15 min at 95 °C; five cycles of 10 s at 95 °C and 30 s at 58 °C; and 45 cycles of 10 s at 95 °C, 30 s at 58 °C, and 15 s at 72 °C. The assay results were interpreted as positive for EML4-ALK according to the manufacturer’s instructions. A positive result was defined as a threshold cycle (Ct) value of less than 40, and the internal control was defined as a Ct value less than 36. The assay results were regarded as invalid if the assays for the EML4-ALK fusion gene and internal control all showed simultaneously negative results. When an invalid result for RT-PCR was obtained, the assay was repeated using newly synthesized cDNA (Fig. 2b).

Treatment responses for chemotherapy and ALK TKIs were evaluated according to the revised RECIST version 1.1 (Eisenhauer et al. 2009). Progression-free survival (PFS) was defined as the time (in months) from the first date of chemotherapy or ALK TKI treatment until the date of objective progression of disease or death from any cause.

We collected clinical information of enrolled patients at diagnosis and during the chemotherapy or ALK TKI treatment. All data are expressed as means ± standard deviations (SD), median (range), or numbers with percentages. Intergroup comparisons were performed using the Mann–Whitney U test for continuous variables and Pearson’s χ² test or Fisher’s exact test for categorical variables. Survival times were estimated for each group using the Kaplan–Meier method. Statistical analysis was performed with IBM SPSS statistics version 25 (SPSS, Inc., an IBM Company, Chicago, IL, USA), and differences with p values of less than 0.05 were considered as statistically significant.

Excluding patients who did not have available tissue specimens or refused to provide blood samples, 33 patients with FISH-positive results and 28 patients with FISH-negative results were finally enrolled. RNA extraction was performed in each group according to the experimental protocol, and cases with inadequate tissue for analysis were excluded (Fig. 1). The mean age of enrolled patients was 63.7 (± 10.7), and the FISH-positive subgroup had younger patients than the FISH-negative group. Moreover, the FISH-positive subgroup had more patients with brain metastasis and brain radiation therapy. There were no differences in sex, smoking history, EGFR mutation status, or response to pemetrexed treatment between the two subgroups (Table 1).

The validation data compared with FISH for tumor tissues are described in Table 2, and representative images for FISH and RT-PCR are shown in Fig. 2. All experimental data about RT-PCR positivity according to biopsy resources are displayed in Fig. 3. RT-PCR using FFPE tissues showed 54.5% sensitivity, 78.6% specificity, and 75.5% accuracy. Liquid biopsy (plasma or platelets) had higher sensitivity (78.8%), specificity (89.3%) and accuracy (83.6%). Platelets showed slightly higher sensitivity than plasma (Table 2).

A comparison of molecular and clinical characteristics in FISH-positive patients is shown in Table 3. Median proportions of positive cells in FISH were likely to be higher in subgroups of liquid biopsy-positive results (plasma, 20.0% versus 15.0%, p = 0.082; platelets 20.0% versus 15.0%, p = 0.084). All blood samples of liquid biopsy-positive subgroups were collected after initiation of systemic treatment, regardless of the types (chemotherapy or ALK TKI). However, platelet showed more liquid biopsy-negative result when the patients were treated with ALK TKI rather than chemotherapy (85.7% versus 14.3%, p = 0.077). Higher positivity for liquid biopsy was shown in the subgroup of delayed blood sampling since diagnosis more than 6 months (plasma, 85.7%, p = 0.106; platelets, 87.0%, p = 0.036). In a subgroup who had a blood sampling within 6 months, 62.5% of samples were drawn after start of systemic treatment (Table 4). In another subgroup who had a blood sampling after 6 months, all samples were drawn after start of systemic treatment. Positive rate of liquid biopsy was relatively higher in patients who were progressed after chemotherapy and preparing ALK TKI treatment than in those who were in ALK TKI treatment.

Among 26 patients treated with crizotinib, the platelet-positive subgroup showed longer duration of treatment (median, 7.2 versus 1.5 months, p = 0.090), longer median PFS [5.7 months, 95% confidence interval (CI) 0.0–16.7 versus 1.7 months, 95% CI 0.5–3.0; p = 0.028], a higher overall response rate (70.6% versus 11.1%, p = 0.011), and a higher disease control rate (88.2% versus 44.4%, p = 0.028) than the platelet-negative subgroup.

Among 26 patients treated with crizotinib, we performed serial collection of blood samples in 12 patients. Four patients were initially positive and eight patients were negative for liquid biopsy. Dynamic change of ALK status in liquid biopsy was as follows: sustained positive (n = 1), positive conversion (n = 5), sustained negative (n = 3), negative conversion (n = 3). Most of the blood samplings were performed earlier from tissue diagnosis (median, 1.5 months, range 0.0–21.2) in patients available for serial monitoring than in those without serial collection (median, 14.5 months, range 0.0–120.0). Overall response to crizotinib of 12 patients was 50.0% (6/12) and disease control rate was 91.7% (11/12). Median PFS was 5.4 months (95% CI 5.3–5.6), and patients with positive results (sustained positive and positive conversion) showed a numerically shorter median PFS (5.4 months, 95% CI 1.9–8.8) that those with negative result (sustained negative and negative conversion, 10.5 months, 95% CI 0.0–21.4, p = 0.174). Representative cases are presented in Fig. 4.

Figure 4a details a case of a 52-year-old woman with stage IV ALK-positive NSCLC who had multiple metastases to the right pleura, both adrenal glands, and the brain. She showed continuous positivity in liquid biopsy during crizotinib treatment, but negative conversion was developed after debulking surgery for colon metastasis followed by alectinib treatment. In this case, the positivity of liquid biopsy was correlated in order with the clinical course.

Figure 4b details a case of a 67-year-old woman with lung-to-lung metastases. Her tumor was ALK-positive NSCLC and showed dramatic regression after crizotinib treatment. Liquid biopsy showed continuous negativity after the initial positive result only in platelets, and the results were correlated with findings of computed tomography scanning as a partial remission without recurrence.

Figure 4c details a case of a 78-year-old woman who was initially diagnosed with stage IIIB ALK-positive NSCLC. Liquid biopsy was negative after concurrent chemoradiotherapy (CCRT). However, upon completion of CCRT, right-sided hemiparesis was developed with multiple brain metastases. After receiving stereotactic brain radiotherapy (gamma-knife radiosurgery), she was treated with crizotinib, and negativity of liquid biopsy was maintained until 3 months before brain recurrence. After positive conversion, positive results of liquid biopsy were continued, along with recurrent brain metastasis; whereas, the intrathoracic lesion still had no evidence of recurrence.