Introduction
Lung cancer is one the most common types of cancer, and the biggest cause of cancer deaths in Wales. It has a very poor prognosis, but survival rates can vary widely, depending on factors such as stage of the cancer at presentation and mode of presentation, such as emergency admission. A prognostic factor is any measure (for example, an aspect of the cancer or a characteristic of the patient) that is associated with the risk of future health outcomes in those with existing disease. In clinical practice, they play an important role in developing a treatment plan, and for providing reliable information on prognosis or survival. Prognostic factor research is conducted in order to discover and evaluate factors that might be useful as modifiable targets for interventions to improve outcomes, building blocks for prognostic models, or predictors of differential treatment response.
The overall aim of the review was to identify prognostic and predictive factors associated with lung cancer survival in the published literature, in order to inform a multivariate model of Welsh data that will be conducted by the Welsh Cancer Intelligence and Surveillance Unit (WCISU). The multivariate modelling will be used to determine modifiable factors that can inform initiatives to potentially improve lung cancer survival in Wales.
Methods
In view of the large volume of lung cancer research, we took a pragmatic approach based on a systematic overview of reviews. It included an initial mapping review, which was based on the titles and abstracts only, to identify all the candidate prognostic factors and the type of prognostic research that has been conducted. A subsequent, more in-depth overview of reviews based on the full-text papers, was then conducted which focused on systematic reviews that aimed to evaluate factors that independently contribute to lung cancer survival. The results provided a summary of whether each factor was associated with survival, the direction of the impact, and whether this was consistent across multiple reviews. Some systematic reviews included a meta-analysis whilst others only presented a narrative synthesis. The initial list of prognostic factors identified during the mapping review was used to identify potentially modifiable factors as defined by two clinicians. These were reviewed in more detail, and where feasible, a summary of the magnitude of effect was developed. The term ‘significant’ was used to indicate statistical significance.
Results
The mapping review included 398 studies, of which 207 systematic reviews that investigated the independent effect of prognostic factors on lung cancer survival were included in the overview of reviews. More than 84 prognostic factors were identified by these reviews, which were grouped under six headers: ‘new’ biomarkers or biological factors not used in routine practice; clinical characteristics or routinely assessed biological variables; tumour characteristics; tumour metabolic activity; patient characteristics; healthcare provider and system factors; and other factors. On the whole the reviews were fairly recent, with only 23 (11%) reviews having conducted their search prior to 2007; 105 (51%) were conducted in the last five years. Most (146/207, 71%) of the reviews limited inclusion to patients with non-small cell lung cancer (NSCLC), whilst only five (2%) reviews focused on SCLC. Thirty two (15%) reviews included the broader category of lung cancer, 21 (10%) covered multiple cancer sites including lung, one review limited inclusion to lung squamous cell carcinoma, one review limited inclusion to lung adenocarcinoma, and one review included multiple
4
primary lung cancers. The most frequently evaluated prognostic factors were novel biomarkers, generally reflecting overexpression or the presence of a mutation. Eighty six novel biomarkers were investigated by 138 reviews, but most (55%) of these were evaluated by a single review. Factors relating to tumour characteristics were also frequently evaluated, with 21 factors investigated in 26 reviews. A further eight reviews evaluated metabolic activity measured using a positron emission tomography - computed tomography (PET-CT) scan. There were also 11 clinical characteristics evaluated in 12 reviews, nine patient characteristics in nine reviews, five factors relating to the heaslatore
lthcare provider in five reviews, and three factors categorised as ‘other’ in four reviews. On the whole the findings of multiple reviews evaluating the same factor were fairly consistent. However, only 16 factors were investigated in more than three reviews: epidermal growth factor receptor (EGFR) (n=11); excision repair cross-complementation 1 (ERCC1) (n=7); Kirsten rat sarcoma viral oncogene homolog (K-RAS) (n=5); microRNA-155 (n=5); survivin (n=5); C-X-C chemokine receptor 4 (CXCR4) (n=4); epithelial cadherin (E-cadherin) (n=4); microRNA-21 (n=4); p53 (n=4); C-reactive protein (CRP) (n=4); histology (n=6); nodal status (n=4); stage (n=6), age (n=3), and gender (n=5).
Twenty of the 207 reviews included in the overview of reviews evaluated fourteen potentially modifiable factors. Eleven (55%) of these reviews were considered as moderate or good quality. A summary of the strength of the association between each modifiable factor and survival is presented. Modifiable factors that were found to be associated with significant improvement in survival were: normal BMI or less weight loss, good performance status (quantitative measure of the patients’ general well-being and activities of daily life), being a non-smoker or never smoked, quitting smoking on or after diagnosis of early stage cancer, lung resection undertaken by thoracic or cardiothoracic surgeon (versus general surgeon), good pre-treatment quality of life, small gross volume tumour, and early stage tumour. Treatment at a hospital with a high procedural volume was associated with significantly better post-operative mortality, but this did not appear to translate to improved overall survival. The patient care being discussed by a multidisciplinary team and having medical insurance were also identified as being significantly associated with improved survival, but the finding for the former was based on a single study and the latter on US studies. Timeliness of care was a significant prognostic factor, but the findings were conflicting, with shorter intervals leading to both increased and decreased survival. This is likely due to a phenomenon called the ‘waiting time paradox’, where patients with rapidly growing or metastatic tumours are likely to present early but have poor outcomes.
Conclusions
The review identified a large volume of published systematic reviews and meta-analyses of prognostic and predictive factors for lung cancer survival. These provide evidence for a long list of prognostic factors, but interestingly few were evaluated by multiple reviews. Where multiple reviews did evaluate the same prognostic factors, their findings, on the whole, were fairly consistent. The review identified several potentially modifiable factors for lung cancer survival, which could be candidates for inclusion in the proposed multivariate analysis by WCISU. In terms of improving cancer outcomes, this body of work could contribute to evidence-based initiatives to improve lung cancer survival in Wales.