RPS 1416a - Chest malignancies: advanced imaging and radiomics

Purpose:

To design a CT reporting system for the classification of sub-solid lung nodules to differentiate early pulmonary adenocarcinoma subtypes (SSNs) in clinical patients.

Methods and materials:

We retrospectively identified 437 SSNs from 2011-2017 with pathological confirmation. SSNs were divided into a training group (N=291) and a testing group (N=146). In multinomial univariable and multivariable logistic regression, CT characteristics were analysed to identify factors discriminating between 3 adenocarcinoma subtypes [pre-invasive lesions (PLs), minimally invasive adenocarcinoma (MIA), and invasive adenocarcinoma (IA)]. The resulting differentiating factors were included in a classification and regression tree (CART) model. Subsequently, a CT reporting system was constructed for SSNs based on the optimised classification model. The classification performance was validated in the testing group.

Results:

The most important differentiating factors based on CT-derived characteristics of sub-solid nodules were visual density (non-solid or part-solid), core (present or absent), diameter of solid component (≤6 mm or >6 mm), and semantic features (vacuole sign, pleural indentation, and vascular invasion). Overall sensitivity, specificity, and diagnostic accuracy of CT reporting system were 89.0% (95%CI: 84.8%-92.4%), 74.6% (95%CI: 70.8%-78.1%), and 79.4% (95%CI: 76.5%-82.0%) for the training group, and 84.9% (95%CI: 78.1%-90.3%), 68.5% (95%CI: 62.8%-73.8%), and 74.0% (95%CI: 69.6%-78.0%) in the testing group, respectively.

Conclusion:

A CT reporting system for sub-solid pulmonary nodules helps to differentiate 3 adenocarcinoma subtypes. The classification tool can be used to assist clinicians in making follow-up recommendations or decisions with regards to surgery in patients with SSNs.

Limitations:

This study is a single-centre retrospective study. The results may be limited to the local population.

Ethics committee approval

n/a

Funding:

Royal Netherlands Academy of Arts and Sciences (PSA_SA_BD_01).

Purpose:

To determine the best CT window width setting to differentiate pre-invasive from invasive sub-solid pulmonary nodules.

Methods and materials:

We retrospectively identified 437 pathologically confirmed SSNs. A solid lesion component on CT images was regarded as a marker of invasiveness (suggestive of MIA or IA). We used a fixed window level (WL) of 35 Hounsfield units and adjusted the window width (WW) until a solid component became visible. This WW was recorded. The best WW cut-off to distinguish invasive and pre-invasive lesions was based on the receiver operating characteristic curve. This WW was defined as the “core” window width. Sub-solid nodules were subsequently categorised as part-solid or non-solid based on the identification of a solid component on the two WW settings, mediastinal window setting (WW/WL, 350/35 HU) and core window setting. The test characteristics were compared.

Results:

88/437 lung lesions were pre-invasive (17 atypical adenomatous hyperplasias and 71 adenocarcinomas in situ) and 349 were invasive (233 minimally invasive adenocarcinomas and 116 invasive adenocarcinomas). The best WW to detect a solid component in SSNs was 1175 HU with AUC of 0.80. Sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy for invasiveness of the SSN were 51.3%, 84.1%, 92.7%, 30.3%, and 57.9% for normal mediastinal window setting and 87.7%, 60.2%, 89.7%, 55.2%, and 82.2%, respectively, for the new core window setting. Of the 233 MIA lesions, 191 (82.0%) were categorised as part-solid based on the core window setting compared to 72 (32.2%) based on the mediastinal window setting.

Conclusion:

A new window setting (WW/WL, 1175/35 HU) outperformed the regular mediastinal window setting (WW/WL, 350/35 HU) for detecting a solid component in SSNs.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

Royal Netherlands Academy of Arts and Sciences-PSA_SA_BD_01.

Purpose:

To compare the capability for postoperative recurrence evaluation among FDG-PET/MRI, whole-body MRI, FDG-PET/CT, and conventional methods based on guidelines in non-small lung cancer (NSCLC) patients.

Methods and materials:

484 consecutive postoperative NSCLC patients (289 men, 195 women; mean age 69 years) prospectively underwent whole-body MRI, integrated PET/CTs, and conventional radiological methods, as well as a follow-up or pathological examinations. All patients were divided into recurrence (n=42) and non-recurrence (n=442) groups based on pathological and follow-up examinations. All co-registered PET/MRIs were generated by means of our proprietary software. The probability of postoperative recurrence in each patient was visually assessed on all methods by means of a 5-point visual scoring system. To compare diagnostic performance among all methods, receiver operating characteristic analyses were performed. The diagnostic accuracy of postoperative recurrence was statistically compared by using McNemar’s test.

Results:

The area under the curves (AUCs) of PET/MRI (AUC=0.99) was significantly larger than others (MRI and PET/CT: AUC=0.97, p<0.05; conventional methods: AUC=0.94, p<0.05). When feasible threshold values were applied, the accuracy of PET/MRI (97.7%) was significantly higher than others (MRI: 96.3%, p<0.05; PET/CT: 94.8%, p<0.05; conventional methods: 90.0%, p<0.05). The accuracy of MRI was significantly higher than PET/CT and conventional methods (p<0.05). In addition, the accuracy of PET/CT was significantly higher than conventional methods (p<0.05).

Conclusion:

FDG-PET/MRI has better the potential for postoperative recurrence evaluation than others in postoperative NSCLC patients.

Limitations:

Qualitative visual assessment for postoperative recurrence.

Ethics committee approval

This prospective study was approved by the institutional review board of Kobe University Hospital, and written informed consent was obtained.

Funding:

Canon Medical Systems, Bayer Pharma and Daiichi Sanky Co., Ltd.

Purpose:

To directly compare the capability for therapeutic outcome prediction among dynamics contrast-enhanced (CE-) perfusion area-detector CT (ADCT) and FDG-PET/CT in small cell lung cancer (SCLC) patients evaluated as limited disease (LD).

Methods and materials:

43 consecutive and pathologically diagnosed SCLC patients assessed as LD underwent PET/CT, dynamic CE-perfusion ADCT, chemoradiotherapy, and follow-up examination. In each patient, therapeutic outcomes were assessed based on RECIST guideline. All patients were divided into 2 groups as follows: responder (CR+PR: n=33) and non-responder (SD+PD: n=10) groups. The total tumour perfusion (TTP) and tumour perfusions from pulmonary (TPP) and systemic (TPS) circulations calculated by the dual-input maximum slope method from dynamic ADCT data and SUV_max were assessed at each targeted lesion. The average value of each index from all targeted lesions was compared between responders and non-responders by Student’s t-test. To compare the differentiation capability in the 2 groups, all indexes as having significant difference were assessed by ROC analysis. Differentiation capability was compared among all indexes as having a significant difference between the 2 groups by McNemar’s test.

Results:

There were significant differences in all indexes except TPP (p<0.05). The area under the curve (AUC) of TPS (AUC=0.92) was significantly larger than that of SUV_max (AUC=0.73, p=004). When applied, each feasible threshold value and the accuracy of TTP (83.7%) and TPS (93.0%) were significantly higher than that of SUV_max (76.7%, p<0.05).

Conclusion:

Dynamic CE-perfusion ADCT has a better potential for predicting the therapeutic outcome than PET/CT in SCLC patients with limited disease.

Limitations:

No comparison of survival.

Ethics committee approval

This prospective study was approved by the institutional review board of Kobe University Hospital and written informed consent was obtained.

Funding:

Canon Medical Systems Corporation.

Purpose:

The accurate staging of tumours invading the mediastinum provides surgeons with critical information for patient selection and surgical planning. Contrast-enhanced CT is currently used for presurgical staging.

The aim of the study is to assess whether cine-MRI improves accuracy in the detection of neoplastic infiltration of critical mediastinal vascular and cardiac structures compared to contrast-enhanced CT, thus possibly expanding the spectrum of patient candidates to radical surgery.

Methods and materials:

From 2008-2018, 32 patients (17M, 25F, mean age 56±14 years) diagnosed with neoplasms invading the mediastinum (lung cancer, n=16; sarcoma, n=7; thymic tumours, n=6; carcinoid, n=1; teratoma, n=1; germ cell tumours, n=1) underwent both contrast-enhanced CT and ECG-gated cine-MRI for preoperative staging. Imaging studies were reviewed by 2 expert cardio-radiologists.

The presence/site of infiltration of critical (aorta, myocardium, and pulmonary arteries) and non-critical vascular structures were noted and analysed separately. CT and MRI diagnostic performances were assessed using the postoperative anatomopathological report as a reference standard.

Results:

Cine-MRI showed better diagnostic accuracy (90.2% vs 70.5%, p=0.03) in the detection of cardiac, aortic, and pulmonary artery involvement compared to CT. MRI excluded invasion of the myocardium and critical vascular structures in 8/32 (23.5%) patients that were initially considered non-surgical candidates according to CT findings.

Conclusion:

Cine-MRI is more accurate than contrast-enhanced CT for the pre-operative assessment of mediastinal tumours with extensive contact with the heart or the great vessels. In our series, cine-MRI refined resectability in 23.5% of patients considered as non-surgical candidates according to contrast-enhanced CT findings.

Limitations:

The limited number of patients and the unavailability of synchronised-CT studies for comparison with ECG-gated cine-MRI.

Ethics committee approval

ERB approved.

Funding:

No funding was received for this work.

Purpose:

To identify 18F-FDG-PET-CT quantitative imaging markers for survival in patients with advanced non-small cell lung cancer (NSCLC) treated with PD-1/PD-L1 inhibitors.

Methods and materials:

This prospective, single-centre study included 40 patients with NSCLC who underwent 18F-FDG-PET-CT before and 8-12 weeks after the start of treatment with a PD-1/PD-L1 inhibitor. We semi-automatically extracted the following parameters: SUV_Peak, TLG (total lesion glycolysis), MTV (metabolic tumour volume), SLR (the spleen-to-Liver ratio), and the bone marrow-to-liver ratio at baseline and the follow-up scan. Each parameter at baseline was dichotomised using the median and percent changes of the parameter were dichotomised using ≤50% reduction as the threshold, respectively. We assessed progression-free survival (PFS) and overall survival using the Kaplan-Meier test and Cox regression analysis.

Results:

In the univariate analysis, high baseline TLG (hazard ratio (HR): HR=3.05), MTV (HR=3.05), SLR (HR=2.70) as well as changes in TLG (HR=5.84), SUV_peak (HR=7.40) and MTV (HR=3.70) were associated with increased mortality (p<0.05). In the multivariate analysis, only the high baseline SLR (adjusted aHR=3.51, p=0.008) and changes in TLG (aHR= 6.90, p<0.001) were independently associated with increased mortality. Similarly, these parameters were associated with reduced PFS, while in the multivariate analysis, only changes in TLG were independently associated with reduced PFS (aHR=3.54, p=0.003).

Conclusion:

Changes in the metabolically active tumour burden correlate with PFS and overall survival. Increased metabolic activity of the spleen at the start of the therapy is an independent risk factor for mortality.

Limitations:

The small sample size and single-centre study. Different PD-1 and PD-1L inhibitors.

Ethics committee approval

Ethics committee approval obtained.

Funding:

Oesterreichische Nationalbank (grant 16886), “Fond für interdisziplinäre Krebsforschung der Stadt Wien,” and the Theodor Koerner Fund.

Purpose:

To assess the relationship of FDG uptake and iodine-related parameters in non-small cell lung cancer (NSCLC) with a focus on the therapy response monitoring and prediction.

Methods and materials:

Patients (n=45) with confirmed NSCLC stage IIIB and IV who were not qualified for concomitant chemoradiotherapy were included in the study. FDG-PET/CT using single-source dual-energy CT (DE-CT) was performed for staging and early follow-up (after the 2nd cycle of chemotherapy). The correlation of FDG uptake and iodine-related values was assessed and compared with the therapy response.

Results:

A strong correlation was found between volumetric FDG parameters (MTV=metabolic tumour volume and TLG=total lesion glycolysis) and the total iodine uptake (mg) using the Spearman correlation coefficient in staging (r=0.0.874, 0.894) and follow-up (r=0.935, 0.934). We also found a significant correlation of change in these values after 2 cycles of therapy. In the prediction analysis, we proved a significant correlation of iodine uptake, MTV, and TLG with the outcome, and iodine uptake was found as a possible strong predictor (r=0.711; p<0.0001). The change in iodine uptake after chemotherapy correlated with an early therapy effect (r=0.659; p<0.0001).

Conclusion:

Our results suggest possible interchangeability in functional assessment. Iodine uptake and volume metabolic parameters were found as predictors of early therapy effect and could be used for a personal approach in therapy conducting.

Limitations:

The number of patients and short follow-up period.

Ethics committee approval

The study was approved by institutional ethics committee and patents signed informed consent.

Funding:

The Ministry of Health of the Czech Republic, grant nr. 17-30748A.

Purpose:

To evaluate early radiological findings that help to discriminate between pseudoprogression and true disease progression in patients with advanced lung cancer treated with immunotherapy.

Methods and materials:

We retrospectively evaluated CT scans of 62 patients with non-small cell lung cancer (NSCLC) treated with PD-1 inhibitor using the iRECIST criteria and considering the invasion of adjacent healthy tissues as a possible predictor of true progression.

Patients considered in unconfirmed progressive disease (iUPD) underwent a second CT scan 4-8 weeks after suspected progression.

Results:

During follow-up, 39 iUPDs were documented; 23 cases after the first cycle of immunotherapy and 16 later, during treatment.

The patients presented different CT scan findings: 4 with new lesions, 28 with an increase in the volume of preexisting lesions, and 7 cases of mixed response were documented.

Local aggressive tumour behaviour with an invasion of nearby healthy structures was documented in 5 patients.

A restaging CT scan performed 4-8 weeks later confirmed true progression (immune confirmed progressive disease, iCPD) in 36 cases, thus leading to a suspension of immunotherapy. All cases of local aggressive behaviour were confirmed as true progression. Only 3 cases of true pseudoprogression were reported (4.8%).

Conclusion:

In our experience, the rate of pseudoprogression did not exceed 5%, which is in line with the literature. In clinical practice, considering local cancer aggressiveness could be a useful criterion to allow a more confident differential diagnosis between pseudo and real progression in addition to dimensional iRECIST criteria. This observation, that should be validated in a larger population, could help in identifying a subgroup of patients at a higher risk of real progression, which might be a useful element for clinical decision making.

Limitations:

The poor sample, limited to advanced lung cancer.

Ethics committee approval

n/a

Funding:

No funding was received for this work.