0115 966 7955 Today's Opening Times 10:00 - 20:00 (BST)

Lung Cancer (NSCLC) Adjusted Prognosis Analysis (APA)

Published: Last Edited:

Disclaimer: This essay has been submitted by a student. This is not an example of the work written by our professional essay writers. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.


Establishment of adjusted prognosis analysis (APA) for initially diagnosed non-small cell lung cancer (NSCLC) with brain metastases in Sun Yat-Sen University Cancer Center

Methods. This retrospective analysis included 837 patients with initially diagnosed brain metastasis from NSCLC in Sun Yat-Sen University Cancer Center, China, from 1994 to 2011. Cox proportion hazard model was used to analyse survival. Factors analyzed included patient characteristics, disease characteristics and treatments. The APA was built according to the number score derived from the hazard ratio (HR) of each independent prognostic variable. The prognostic score of each patient was calculated by totaling up the scores of each independent variable.

Results. Six factors were independently associated with the survival time: KPS, age, smoking history, intracranial metastasis local treatment, received TKI treatment, and received Chemotherapy. APA model 1 was built and the 837 patients were well stratified into low (score 0-2), moderate (score 3-5) and high risk group (score 6-7). Median survival time of three groups were 16.6, 10.3 and 5.2 months, respectively (p<0.001).

A subgroup analysis was made for 221 patients with indentified EGFR status to built APA model 2. EGFR mutation status was an independently prognosis factor which replaced smoking history in model 2.

Keywords: Brain metastases, non-small cell lung cancer, survival model, prognosis, EGFR


Brain metastases occur in up to 40% of cancer patients and become a major cause of mortality and morbidity1. The majority of brain metastases originate from primary cancers in the lung (40-50%), breast (15-25%) and melanoma (5-20%)2,3. During the last decade, the incidence of brain metastases was increasing as a result of improvements in the diagnosis and systemic treatment of extracranial disease4. Approximately 10-20% of NSCLC patients present with synchronous brain metastases at the time of diagnosis5. Brain metastasis are associated with a poor prognosisdespite treatment with whole brain radiation therapy (WBRT), with a median survival of less than 6 months6. Because of the present of blood-brain barrier (BBB), the role of systemic chemotherapy and molecular targeted agents is controversial.

The most widely used prognostic index for brain metastasis over the last decade is the RPA (recursive partition analysis) index originally described on behalf of Radiation Therapy Oncology Group (RTOG)7. RPA is based on 4 factors (Karnofsky performance score (KPS), age, presence or absence of extracranial metastases, and the control status of the primary tumor), separating patients into 3 different prognostic classes6. The RTOG has recently proposed a new index, GPA (Graded Prognostic Assessment), which was compared to RPA, SIR (Score index for radiosurgery) and BSBM (Basic score for brain metastasis)8.

We conducted a retrospective analysis for newly diagnosed NSCLC with brain metastasis from August 1994 to December 2011 at a single institution. We aim to develop a prognostic model specifically for initially diagnosed NSCLC with brain metastasis based on common clinical factors.

Patients and methods

Inclusion and exclusion criteria

We retrospectively screened patients with BM from initially diagnosed NSCLC and received treatments in Sun Yat-Sen University Cancer Center from August 1994 to December 2011. The eligibility criteria included: â‘  Pathologically confirmed and treatment naive NSCLC; â‘¡ Confirmed brain metastasis by brain MRI; â‘¢ Baseline clinical information was available and complete treatment and follow-up in Sun Yat-Sen University Cancer Center; â‘£ Patients with normal renal, cardiac and liver function. Exclusion criteria were as followed: â‘  Patients presented with symptoms or radiographic findings of leptomeningeal metastases; â‘¡ Patients with other types of malignancy; â‘¢ Women during pregnancy or lactation period. The study was reviewed and approved by the Institutional Review Board and academic committee of Sun Yat-Sen University Cancer Center.

Survival analysis

Overall survival was calculated by the Kaplan–Meier method, from the date of diagnosed NSCLC with BMs to the date of death for any course or the date of the last follow up. OS was analyzed using the Kaplan-Meier method and was compared using the log-rank test. Univariate and multivariable analyses were performed using the Cox’s proportional hazards model. Factors that were considered for survival analyses included patient characteristics (gender, age group, performance status , smoking history), disease characteristics (histology, extracranial distant metastases, number of brain metastatic lesion, Intracranial hypertension, EGFR mutation status) and treatments (thoracic local treatment, intracranial metastasis local treatment, tyrosine kinase inhibitors treatment, chemotherapy). We defined overall survival as the time from the first date diagnosis of brain metastasis to the time of death. Karnofsky Performance Scores were determined at the time of diagnosis of brain metastasis. Follow up date were recorded until September 2013 or death. All statistical tests were two-tailed and P < 0.05 was considered significantly. Statistical analysis was performed by SPSS 16.0 (SPSS Inc., Chicago, IL, USA).

APA model establishment

The hazard ratio (HR) of each independent prognostic variable was changed into an integral number to build the APA (adjust prognosis analyses) model for newly diagnosed brain metastases from NSCLC.


Patient characteristics

837 patients with newly diagnosed BMs from NSCLC were eligible for analysis. All patients were Chinese ethnicity with a male predominance (68.8%). About half of the patients (50.2%) were ever-smokers. Only 221 patients (26.4%) had identified EGFR type. The patient characteristics were listed in Table 1.

Treatment Outcomes

Of all patients, 366 patients (43.6%) received WBRT, 41 patients (4.3%) received surgery for brain metastasis, and 28 patients (3.3%) received SRS. One hundred and nineteen patients (14.2%) received thoracic local treatment, which included surgery and radiation. Only 20.7% patients were treated with TKIs (including Gefitinib, Erlotinib and Icotinib). An overwhelming majority (91.8%) of the patients received chemotherapy. And 58.1% patients only received one regimen chemotherapy. 20.2% patients received chemotherapy and target therapy.

Treatment pattern were shown on Table 2.

Survival Outcome

Overall, 670 deaths were recorded (80%), with median OS in the whole population equal to 13.4 months. The 6 months and 12 months survival rate were 80.8% and 49.7%, respectively. Univariate and Multivariate Analysis

The univariate analysis was performed on 13 different variables (Table 2) to evaluate their potential value on overall survival of the whole group. Seven factors had significantly impacted survival, which were KPS, age≥60, smoking history, EGFR status, TKI therapy, local therapy for BM and chemotherapy.

Because EGFR mutation type was known for only 26.4% patients, we excluded it from multivariate analysis for whole group. The other 6 factors were included in the multivariate analysis. Table 3 summarizes the result of the multivariate analysis for survival. Six factors were independently associated with shorter overall survival: KPS≤70 (HR=2.131, P<0.001), age≥60 (HR=1.231, P=0.008), smoking history (HR=1.176, P=0.034), not received TKI therapy (HR=1.730, P<0.001), not received local therapy for BM (HR=1.350, P<0.001) and not received chemotherapy (HR=1.387, P=0.017).

RPA and GPA Analyses

According to RPA, the patients were only stratified into two groups: classes 2 (701 patients) and class 3 (136 patients). In class 2 (KPS≥70), 701 patients’ median survival time (MST) was 14.7 months. In class 3 (KPS<70), 136 patients’ median survival time was 5.7 months. MST was significantly different between class 2 and class 3 (14.7 VS. 5.7 months, P<0.001, Figure 1a). According to GPA, the whole group patients were stratified into four group, and the middle survival time were 11.6 months (GPA 0-1.0, n=146), 13.5 months (GPA 1.5-2.5, n=504), 15.3 months (GPA 3.0, n=129), and 11.7 months (GPA 3.5-4.0, n=58), respectively. The survival curves of GPA for patients was shown in Figure 1b. The GPA 3.0 group had the longest survival time. The survival curves of GPA were crossover.

Establishment of APA model 1

Six factors were classified by two category (Table 4a). According to the hazard ratio (HR), our model assigned 2 points for KPS≤70 and 1 point for the other factors. The score for each patient was calculated by summing the points of six independent factor. The maximum score is 7. 837 patients were classed into three groups according to different prognosis, which including low risk groups (score 0 to 2, n=420 patients), intermediate risk groups (score 3-5, n=337 patients), and high risk groups (score 6 to 7, n=80 patients). The median survivals of the three groups were 16.6, 10.3 and 5.2 months, respectively (P<0.001). The OS curves of different group were distinctly separated from each other (Figure 2), whereas in RPA the OS curves only separated into two lines and in GPA the OS curves were crossover. Our study showed that APA model 1 was superior prognostic model than RPA and GPA for newly diagnosed NSCLC with brain metastasis.

Subgroup Analysis for EGFR-known Patients and Establishment of APA model 2

Subgroup analysis were made for 221 patients with identified EGFR status. Among this subgroup, 68 patients (30.1%) had EGFR mutation, the others 153 patients (69.9%) were wild type . The subgroup patient characteristics were shown in supplement Table 1. Univariate and multivariate analysis using COX regression analysis for the 221 patients also showed 6 factor which had significant correlation with overall survival (supplement Table 2). Where in this subgroup analysis, EGFR became an independent prognosis factor by replacing smoking history. The median OS for non-smokers was numerically but not statistically longer compared with smokers (17.2 months VS. 14.2 months, P=0.77). According to the results of subgroup analysis, we established APA model 2 for the EGFR-known NSCLC patients with brain metastasis in which the smoking history in APA model 1 was replaced by EGFR wild type. Basing on the same method mentioned in APA model 1, the total scores were calculated for 221 patients. Three risk stratification groups were obtained basing on APA model 2: the low risk group (total score=0-2) which included 128 patients; the intermediate risk group (total score=3-5) which included 88 patients; and the high risk group (total score=6-7) which included only 5 patients. The median survival time of the 3 groups were 21.1, 14.2 and 3.9 months, respectively (P<0.001). The OS curves of subgroup known EGFR patients were shown in fig 3. 3


We have conducted the first study of building a survival model specially for newly diagnosed BM from NSCLC and establishing the Adjusted Prognosis Analysis (APA). Up to now, six published prognostic indices have been used to establish overall survival for brain metastasis6-12. RPA and GPA are the most popular. Compared with RPA and GPA, our models have some difference, which is shown in table 3. In all prognosis indices including our survival model, performance status and age are the prognosis factors. However, In our study, the number of BM and extracranial metastases came out to be the prognostic factors with no significance.

Until now, whether the number of BM as prognosis factors were still controversial. GPA which about 20% patients were recorded having more than 3 brain metastasis8, is not in accordance with our model which 318 patients (38.0%) had >3 brain metastasis. In most of prognosis indices, extracranial metastases was one of the significant prognostic factor, but it didn't include in our survival model. The reason may be that most patients who were analyses in other models, received local treatment for brain metastases. Whereas in our analysis, 48% patients didn't receive local treatment for brain lesion so the main course of death was the progression of extracranial metastases. But. The difference of treatment pattern maybe the most important reason for excluding extracranial metastases from our model.

The smoking history and EGFR mutation type were included in our survival model, which has been the most important difference with the former models. Kogure Y. et al. reported that higher pack-years and shorter period since cessation of smoking were significantly associated with poorer survival in NSCLC13. And Kawaguchi T. et al14 found that never-smoking status conferred significant improved OS for Japanese and a trend for improved OS for Caucasian NSCLC patients. However, a few study focused on the prognosis of BM with cigarette smoking. In Wu, Y. L.'s research, which confirmed that non-smoker had a better survival for smoker (21.3 months VS.5.7 months, p=0.02) in a small sample size of BM patients15. Our study revealed that never-smoking status significantly prolonged the survival time (14.7 months VS.11.5 months, p=0.002) in newly diagnosis NSCLC with BM. Therefore, it is rational that the smoking status stood out to be the independent prognostic factors in our survival model.

EGFR mutation status has been reported as an important prognostic factor for NSCLC. Stanic K reported that the overall survival in EGFR positive patients with BM at diagnosis was longer than in EGFR negative patients (12.6 vs. 6.8 months, p=0.005)16. In EGFR-known subgroup analysis, we also found that EGFR mutation type was an independent positive prognostic factors for survival outcome of BM patients from NSCLC. EGFR type was included into APA 2. Interestingly, active tobacco smoking has been negatively related to the incidence of EGFR mutations17,18, and when EGFR mutations were considered into analysis, smoking history were no longer the independent factors. Although chemotherapy or EGFR-TKI plays an definitely treatment role for NSCLC, it remains controversial whether chemotherapy or EGFR-TKI could improve the survival of NSCLC with BM patients. Sanchez de Cos, J. et al. reported that the strongest favorable prognostic factor was active treatment of both the primary tumor (surgery, chemotherapy and/or thoracic radiotherapy) and brain metastasis (neurosurgery and/or whole brain radiotherapy) for NSCLC with brain metastases19.

Our study showed that the APA model 1 was feasible for those NSCLC BM patients with unknown EGFR genotype, whereas considering molecular biology and EGFR genotype, the APA model 2 seemed more appropriate for BM patients from NSCLC with known EGFR status. However, there were limitations in our study. All patients came from a single center.

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Request Removal

If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please click on the link below to request removal:

More from UK Essays

We can help with your essay
Find out more
Build Time: 0.0088 Seconds