Our study demonstrated that EGFR mutation positivity is predictive for better overall survival. The median overall survival improved from 0.79 years in EGFR mutation-negative group to 2.81 years in EGFR mutation-positive group. This finding supports our earlier New Zealand study, which showed gains in overall survival in EGFR mutation-positive group compared to EGFR mutation-negative group and the group of patients who were untested for EGFR mutation 33. The current study adds better understanding of determinants of overall survival in non-squamous NSCLC patients with different EGFR mutation status.
We found the extent of the disease (presence of metastasis or not) was the other major determinant of overall survival in addition to EGFR mutation status. These two factors are largely independent, so median overall survival increases from 0.38 years in metastatic EGFR-negative, to 1.39 years in metastatic EGFR-positive, 2.31 years in non-metastatic EGFR-negative, and 4.28 years in non-metastatic EGFR-positive cancers.
Within these four groups of cancers (based on metastasis status and EGFR mutation status), the effects of most other factors are similar. Performance status showed large and significant effects, which are similar in all four groups. Overall survival is diminished in those over 80 years, with similar effects in the four groups, while at ages up to 79 age has little effect on overall survival. There is some evidence that sex may modify the EGFR mutation effect on overall survival; females had significantly better overall survival for EGFR-positive disease with no metastasis, but equivalent overall survival for EGFR-negative metastatic disease. However, females had worse overall survival for EGFR-positive disease, either metastatic or not, although the differences in overall survival were not statistically significant. With regard to ethnicity, no differences were seen for non-metastatic disease, but for metastatic disease, both EGFR-negative and EGFR-positive, overall survival was significantly lower in Māori patients.
Our finding of age at diagnosis, smoking status and performance status being significantly associated with survival in both EGFR mutation groups was consistent with that of a Taiwanese study of overall survival of advanced NSCLC patients 21. However, in the Taiwanese study, being female was associated with better survival in both EGFR mutation groups.
Ethnicity was studied specifically in this New Zealand study. Māori, the indigenous people in New Zealand, and Pacific people have a higher incidence of lung cancer and poorer survival, compared to the New Zealand European population 34. This study found that the survival of Māori patients was significantly lower in the metastatic group, and it was more profound in the EGFR mutation-positive group. The survival of Pacific patients was also lower than that of New Zealand Europeans, but this association was non-significant. EGFR mutation is less prevalent in Māori patients and the benefit of treatment with EGFR-TKI appears to be more limited. However, on a population basis, as lung cancer incidence in total is higher in Māori, the numbers of mutation-positive Māori patients who may benefit, expressed as a proportion of the population, could be higher than that of mutation-positive non-Māori patients. Thus, the survival disparity between different ethnic groups indicates an area to address in improving overall survival of NSCLC patients.
The primary site of tumour was an important predictor of overall survival in the patients without metastasis in our study. It supports the findings of previous studies, which reported that upper lobe tumours were associated with better survival compared to those located in other sites of the lung 17,35. Our study also revealed an association between tumours in the main bronchus and poorer overall survival compared to the upper lobe tumours. For this association between tumour site and overall survival, the EGFR mutation status did not seem to play a significant role.
Our analysis adds knowledge to existing evidence of survival benefits in EGFR mutation-positive NSCLC patients. A previous meta-analysis of 19 published studies concluded that EGFR mutation itself was not a prognostic indicator, based on their finding of similar post-operative disease free survival in resected NSCLC patients 22. Yet, in advanced stage patients who were treated with EGFR-TKIs, EGFR mutation was predictive for progression free survival 13, implying that the survival benefits are attributable to treatment following the detection of EGFR mutation. Conversely, previous major randomised controlled clinical trials showed no significant differences in overall survival between EGFR-TKI therapy versus chemotherapy in EGFR mutation-positive patients 14,33. Our study confirms the improved overall survival in EGFR mutation-positive patients and reports the effects of various predictive factors on overall survival.
Our study included a total population-based cohort of non-squamous NSCLC in northern region of New Zealand, which contributes about 40% of New Zealand population. Our data had less than 10% missing data on smoking status and performance status. Yet, we considered possible limitations of our study. Although our cohort included a considerably large number of patients (n = 1534), the EGFR mutation was positive in 20% of patients (n = 310), which limits the power of subgroup analyses. The EGFR mutation information was based on the patient population tested for EGFR mutation where testing was incomplete – the testing rate was 65% among eligible non-squamous NSCLC based on the same region data in 2014, and attributable to patient selection bias 36. We plan to assess the changes when testing becomes more complete in the future.
We conducted Cox regression analyses separately based on EGFR mutation status. We performed further separate analyses based on metastasis status to improve compliance with proportional hazards assumptions. We kept the same factors in both EGFR mutation-positive and -negative models for comparison purposes. EGFR mutation has been proven to be more common in female, Asian, non-smoker and adenocarcinoma lung cancer patients internationally 37−39. In New Zealand, our previous study showed that EGFR mutation was more prevalent in female, in Asian and Pacifica compared to New Zealand European, and in non-smokers compared to current smokers 33,36. Considering statistical significance of the factors and interactions between factors may be necessary for developing survival predictive models in the future.