Introduction

Lung cancer is the leading cause of cancer-related death worldwide [1], despite recent advancements in screening guidelines [2] and treatment options. Non-small cell lung cancer (NSCLC) comprises approximately 80% of all lung cancers, and most patients present with locally advanced to advanced disease. The treatment landscape for patients with advanced disease has changed considerably in recent years, with the incorporation of targeted therapy and immunotherapy into the standard of care [3,4,5,6,7,8]. Outcomes for patients with advanced disease, however, remain poor overall.

Much effort has been placed on identifying prognostic factors in NSCLC, with disease stage and performance status being among the most well-described [9]. Positron emission tomography with 18-fluorodeoxyglucose (FDG PET) is a key procedure in the workup and staging of NSCLC. Metrics extracted from PET have also emerged as potential prognostic tools. The standardized uptake value (SUV) is defined as the ratio of tissue radioactivity concentration and administered dose of radioactivity. The maximum SUV (SUVmax) is often used to quantify a lesion’s metabolic activity, however, it has not been proven to be a valuable prognostic marker. Metabolic tumor volume (MTV) and total lesion glycolysis (TLG), on the other hand, are volumetric parameters that have shown promise in predicting clinical outcomes. MTV is the sum of the volume of all voxels with an SUV above a pre-determined threshold value. TLG is calculated by multiplying the MTV and the average SUV, or SUVmean, combining both volumetric and metabolic information.

In 2015, Im et al. published a meta-analysis demonstrating the prognostic value of these parameters across all stages of non-small cell lung cancer [10]. High MTV and high TLG were associated with increased risk of all-cause mortality, with hazard ratios of 2.31 and 2.43, respectively. Our objective was to perform an updated meta-analysis of the published literature characterizing the prognostic value of pre-treatment, volume-based FDG-PET metrics in NSCLC, specifically focusing on patients with advanced disease and considering recent treatment advancements.

Materials and methods

Data search and study selection

We conducted a systematic PubMed search to identify studies describing the prognostic value of baseline volume-based PET metrics, including total MTV and/or TLG in patients with advanced NSCLC. We searched for English language studies using the search terms “positron emission tomography,” “advanced lung cancer,” “metastatic lung cancer,” “stage IV lung cancer,” and “volume.” We also searched articles related or similar to those yielded in our initial search. We included publications with advanced NSCLC, baseline FDG-PET obtained prior to initiation of first-line systemic therapy, and measurement of total body tumor burden via MTV and/or TLG. We did not include studies where only the primary tumor was measured. Since we wanted to focus on patients with advanced, or metastatic, disease, for whom the standard of care therapy has recently changed, we chose a cutoff of at least 50%, or a majority of study patients having Stage IV disease, for those studies which included a range of early and late stage NSCLC.

Data extraction and statistical analysis

We recorded study characteristics, including first author, location and year of publication, number of patients, percentage of patients having Stage IV disease, forms of treatment administered, PET techniques, and cutoffs used for high MTV and TLG. The clinical endpoints examined were progression-free survival (PFS) and overall survival (OS). Hazard ratios for PFS and OS were extracted directly from the original reports when available, or estimated indirectly from survival curves, as previously suggested by Parmar et al. [11], using customized scripts in Matlab (The Mathworks, Natick, MA). Briefly, our customized scripts function by digitizing published survival curves and inferring the timing of each patient’s clinical event or censoring. Cox proportional hazards models are then fit to the derived patient-level data. Inverse variance meta-analyses using random effects models were performed to assess associations between PET metrics and clinical outcomes. The Q Test was used to identify potential outliers within the data set.

Results

Our electronic search results yielded 416 records. We eliminated 358 entries upon initial screening and identified 58 abstracts to assess further for eligibility. Fourteen full-text articles met our eligibility criteria. We also reviewed articles that were listed as similar to these fourteen entries and upon further investigation, we identified an additional 7 publications to include for analysis. One article was subsequently found to have been retracted and was excluded. Another 7 articles had either duplicate or insufficient data. Thirteen publications were included in our final analysis (Fig. 1), incorporating 1047 patients [12,13,14,15,16,17,18,19,20,21,22,23,24].

Fig. 1
figure 1

Search strategy and study inclusion

Full size image

The percentage of patients with Stage IV disease ranged from 67 to 100%. There were 3 studies we included [21, 23, 24] which did not provide a breakdown of stage, but these articles specifically stated patients with advanced disease were studied. Two of these articles specified that patients with at least Stage IIIB were included [23, 24]. Patients from at least nine studies received chemotherapy and patients from at least 4 studies were treated with targeted therapy. Patients in one study with PD-L1 scores of at least 50% were treated with first-line immunotherapy (Table 1). The majority of the included publications described methods of MTV and TLG delineation, and PET/CT techniques including patient fasting time prior to imaging, injected activity, uptake time, scan time, and reconstruction and attenuation correction details (Table 2).

Table 1 Study characteristics
Full size table
Table 2 PET Parameters
Full size table

The median cutoff used to define high MTV across studies was 93 cm3. Random effects models demonstrated that high MTV is significantly associated with inferior PFS (Fig. 2), with a hazard ratio (HR) of 2.97 (95% CI 2.21–4.00, p < 0.001), as well as inferior OS (Fig. 3), with a HR of 2.73 (95% CI 2.18–3.41, p < 0.001). Similar findings were seen with regards to TLG. The median cutoff used to define high TLG across studies was 444 cm3. Random effects models demonstrated that high TLG is significantly associated with inferior PFS (Fig. 4), with a HR of 2.13 (95% CI 1.56–2.91, p < 0.001) as well as inferior OS (Fig. 5), with a HR of 2.06 (95% CI 1.75–2.44, p < 0.001).

Fig. 2
figure 2

Association between MTV &PFS

Full size image
Fig. 3
figure 3

Association between MTV & OS

Full size image
Fig. 4
figure 4

Association between TLG & PFS

Full size image
Fig. 5
figure 5

Association between TLG & OS

Full size image

Discussion

In our analysis, high MTV and high TLG were significantly associated with worse PFS and OS outcomes in patients with advanced NSCLC. Several studies have been published in recent years examining the prognostic value of pre-treatment PET/CT in advanced NSCLC across multiple lines of therapy [25,26,27,28,29]. This meta-analysis establishes its value prior to first line therapy, thus providing a useful tool to help guide expectations at the time of initial diagnosis.

Our findings largely corroborate a previous meta-analysis published by Im, et al. Unlike the previous study, we focused on patients with advanced NSCLC, mitigating concerns that prior findings could be mediated by associations between disease stage and volumetric measures of disease burden. Additionally, we considered primary tumors and regional and distant metastases as components of patients’ disease burden, while the previous study focused only on primary lung tumors. Of note, future studies may implement machine learning technique to identify more robust PET-based prognostic variables (e.g., textural features [30]).

Of note, only one study meeting criteria for this analysis utilized immunotherapy as first-line treatment for NSCLC. Our group has begun to examine the value of volume-based PET metrics in our own cohort of patients treated with first-line immunotherapy, and has shown that MTV may be an important prognostic factor for these patients with advanced disease [31]. Additional studies that reflect this current standard of care treatment paradigm for NSCLC are sorely needed to confirm the benefit of PET/CT in the modern treatment era.

Stratification factors employed in recent randomized trials for advanced NSCLC include race, ECOG performance status, tumor histological type, region of enrollment, PD-L1 tumor proportion score (TPS), and choice of chemotherapy [3,4,5,6,7,8]. The hazard ratios for PFS and/or OS associated with these factors range from 1.23 to 2.94. MTV and TLG, as evidenced by the HR’s detailed in our study, are arguably more powerful than many of these prognostic tools. These findings support the use of volume-based measures of disease burden as stratification factors to improve clinical trial efficiency.

Aside from volume-based measures, an additional method of assessing tumor burden in advanced disease is to count the number of sites of metastases. The presence of only a few metastatic sites is termed oligometastatic disease and has been associated with a favorable prognosis. Multiple studies are exploring the role of radical local therapy for oligometastatic cancer patients [32,33,34]. Future studies with patient-level data are warranted to examine the relationship between volumetric and count-based measures of disease burden, and the merits of each of these metrics as prognostic factors should be compared.

Limitations to this study include the lack of patient-level PET data and the use of variable definitions of high/low MTV and TLG. Of note, sensitivity analyses performed after removing trials with extreme MTV and TLG cutoffs did not yield meaningful changes to the results of our meta-analyses (data not shown). Future analyses with patient-level PET data would allow more refined characterization of the relationship between disease burden measured on PET and NSCLC prognosis.

Conclusion

Baseline PET metrics are powerful prognostic factors for advanced NSCLC patients who are treated with first line therapy. Future studies are needed to examine the prognostic value of PET metrics for patients who receive first-line immunotherapy.