Virtual Library

Start Your Search

A. Dettrick



Author of

  • +

    P1.06 - Poster Session 1 - Prognostic and Predictive Biomarkers (ID 161)

    • Event: WCLC 2013
    • Type: Poster Session
    • Track: Biology
    • Presentations: 1
    • +

      P1.06-003 - Comparative effectiveness of ddPCR for the detection of EGFR mutations. (ID 357)

      09:30 - 16:30  |  Author(s): A. Dettrick

      • Abstract

      Background
      Selection of EGFR TKIs for lung cancer requires accurate detection of activating mutations. Traditional techniques are limited by small biopsy sizes. We compared droplet digital PCR (ddPCR, Bio-Rad) to Sanger sequencing, mutant-enriched PCR (ME-PCR) and high resolution melt (HRM) PCR.

      Methods
      A comparative effectiveness study was performed on 317 resected NSCLCs with salt extracted gDNA. EGFR exons 19 & 21 Sanger sequencing and (Shigematsu et al., 2005) and HRM/ME-PCR mutation detection (Sriram et al., 2011) was previously reported. ddPCR (Bio-Rad) was performed with competitive allele specific Taqman PCR (CastPCR; Life Technologies); EGFR L858R assay for exon 21 c.2573T>G and c.2572_2572CT>AG; exon 19 deletion assay for 20 common deletions (EGFR_ex19dels_mu, EGFR_6224_mu). 8ng gDNA was tested in 20uL reactions partitioned into 20000 droplets; valid reads contained ≥ 10000 droplets. Controls were 8ng gDNA from mutation positive cell lines (AJCC: H1975, H1650), human female DNA (Promega) and no template controls. QuantaLife (Bio-Rad) calculated Poisson statistics determined allele copy/uL; samples with minimum estimated mutant copy number ≥ 0.15/uL (3 copies/8ng) were “called” positive.

      Results
      Serial dilution of control assay demonstrated detection of template to 40pg input gDNA. 209 (66%) men and 108 (34%) women of mean age 63 years (range 36 to 83) were included. 295 (93%) had smoked and 18 (6%) were never smokers; 4 (1%) were unknown. 1 subject (0.3%) was Asian. pTNM stages were I (47%), II (31%) and III (22%) respectively (6[th] Ed). 171 (54%) were adenocarcinomas, 109 (34%) squamous cell carcinomas, and 37 (12%) other histologies. Figure 1 Exon 19 and 21 ddPCR assays yielded valid results for 300 and 301 samples respectively. ddPCR detected all mutations previously demonstrated by Sanger sequencing (13) and HRM (13) but not ME-PCR (13/15). Mean droplet counts were lower in ddPCR only called (30 droplets/ng) than those samples also called by other methods (3600 droplets/ng; p=0.039). Median percentage tumour and necrosis content of mutation positive samples only by ddPCR were 30% and 0% respectively, identical to those called by other methods.

      Conclusion
      ddPCR identifies mutations detected by Sanger sequencing and HRM. Limitations include nanodrop quantitation of input DNA and data replication is required. This technique demonstrates high sensitivity but limited specificity and requires further validation to examine the significance of low droplet number positive calls. The authors acknowledge the assistance of R Harrison and A Beckhouse, Bio-Rad. Financial support gratefully received from: NHMRC (MD PhD Scholarship), CCQ (MD PhD Scholarship), Cancer Australia, TPCH Foundation, Queensland Health.

  • +

    P3.01 - Poster Session 3 - Cancer Biology (ID 147)

    • Event: WCLC 2013
    • Type: Poster Session
    • Track: Biology
    • Presentations: 1
    • +

      P3.01-011 - Heterogeneity in tumour content and necrosis in primary lung cancers: Implications for molecular analysis (ID 3326)

      09:30 - 16:30  |  Author(s): A. Dettrick

      • Abstract

      Background
      Lung adenocarcinoma (AC) and squamous cell carcinoma (SCC) tumours have a large variance in tumour cell content. This heterogeneity is a concern for genomic studies, as it is difficult to distinguish mutational differences between tumour and non-tumour if low percentage tumour is used for analysis. In addition to this, tumour samples are affected by the amount of necrosis present, as the overall number of viable cells is decreased. We assessed tumour and necrotic content in lung tumour specimens from AC and SCC patients and aimed to identify possible implications for the suitability of these samples in molecular characterisation studies using next generation sequencing technology.

      Methods
      Lung tissue specimens were collected during the period of 1990 to 2013 from patients at The Prince Charles Hospital who consented to donate their surgically resected lung tissues for research. Tissues were macroscopically dissected, snap frozen in liquid nitrogen and stored at -80°C. A tissue section was taken and stained with haematoxylin and eosin (H&E) for two pathologists to independently assess tumour cell and necrotic content. Tumour cell content (TC) in each specimen was scored as percentage of viable cells as seen on the H&E slide, where necrotic content (NC) was recorded as a percentage of the whole slide section. Statistics were calculated using SPSS v21 software. Tumour specimens screened for eligibility to The Cancer Genome Atlas sequencing project are presented here.

      Results
      Tumours from 62 AC and 104 SCC subjects were scored (specimen characteristics in Table 1). Scoring between the two pathologists was highly correlated, with a high intraclass reliability (0.94 and 0.96 for TC and NC respectively).

      Table 1: Clinical and Pathological Characteristics of Specimens
      AC SCC
      Number of Specimens 384 609
      Number of Males/Females 36/26 84/20
      Median Specimens per Subject 4 4
      Range of Specimens per Subject 1-25 1-27
      Median TC 35% 30%
      Range of TC 0-88% 0-90%
      Median NC 0% 6%
      Range of NC 0-90% 0-100%
      Median Age 62 yrs 68 yrs
      Range of Age 45-85 yrs 46-91 yrs
      Median Smoking Pack Years 40 56
      Range of Smoking Pack Years 0-115 0-158
      TC varied from 0-~90% for both subtypes. Comparing AC and SCC, the median TC was higher in AC than SCC (35% vs 30% respectively, p<0.05). NC varied from 0-~100%, but was generally low. The median NC was statistically significantly different between AC and SCC (0% and 6% respectively, p<0.001). TC was weakly correlated with NC (Spearman Rank r = 0.32, p<0.01). There were no clinically important correlations between smoking pack years, gender or age with TC and NC of specimens.

      Conclusion
      Lung AC and SCC specimens are heterogeneous in terms of TC and NC. Therefore, only a small proportion of resected lung cancer specimens meet the criteria required for massively parallel sequencing projects that require high quality tumour DNA and RNA (ie low NC) and relatively low stromal contamination (ie high TC).