Historically, physicians used a combination of performance status (PS), measured using the Eastern Cooperative Oncology Group (ECOG) PS scale, and organ function as determined through blood work to determine, which patients qualified for chemotherapy treatment (32). This approach has been demonstrated to perform poorly when compared to more formal geriatric assessment (33–35). Comprehensive geriatric assessment (CGA) is usually composed of medical, functional, mental, social, and nutritional assessments, as well as explicit assessment of prescription drug use (36). A variety of studies have been completed to evaluate the usefulness of CGA in oncology patients. Systematic reviews of the available evidence show that CGA identifies problems that would otherwise be missed, leads to modifications in treatment plans, and helps predict toxicity from chemotherapy (37–42). Modification of treatment plans occurs in 21–53% of patients, suggesting that oncologists believe the additional information is valuable (42). CGA is also better than physician opinion for identifying frail elderly patients who experience greater toxicity (43). When tested in elderly NSCLC patients, CGA was feasible (44–46). Frail patients also exhibited poorer survival (46). However, when Corre et al. allocated patients to treatment based on CGA, survival was not different between groups, but toxicity was reduced in the arm allocating treatment using CGA (47).

While a CGA can be very useful, it has not become a routine part of oncologic care because it is time and labor intensive. The mean duration of CGA during one prospective study was 80 min/patient (48). Such a time commitment is difficult to undertake in lung cancer patients with metastatic disease, since it may delay patient throughput in clinic and/or delay commencement of treatment, a serious concern when patients have an average life expectancy of 10–12 months (8). In light of these concerns, a number of groups have attempted to shorten the CGA or provide a screening tool. Two groups have published new tools geared toward predicting chemotherapy toxicity and derived from multivariable analyses of CGAs conducted in cancer patients (Table 2) (33, 49). The chemotherapy risk assessment scale for high-age patients (CRASH) is actually composed of two scores, one for hematologic toxicity and another for non-hematologic toxicity (49). Diastolic blood pressure, instrumental activities of daily living (IADL), lactate dehydrogenase, and a proprietary Chemotox score help stratify the likelihood of experiencing Grade 3–4 hematologic toxicity into low (7%), medium-low (23%), medium-high (54%), and high (100%). For non-hematologic adverse events, the predictors are: ECOG PS, mini mental status, mini nutritional assessment, and the Chemotox score. The same predictive categories for non-hematologic toxicity predict risks of 33, 46, 67, and 93% (49). The Chemotox score is a quantitation of the toxicity of chemotherapy regimens that the group developed previously (50, 51). The cohort used for the CRASH score contained 21% lung cancer patients. The Cancer Aging Research Group (CARG) derived another predictive tool (33, 52). The CARG score uses 11 factors to stratify risk: age ≥72, cancer type, standard chemotherapy dosing, polychemotherapy, low hemoglobin, low creatinine clearance, fair or worse hearing, falls, needing help with medications, trouble walking 1 block, and decreased social activity. This score classified lung cancer patients into low (10%), intermediate (40%), or high (60%) risk of Grade 3–5 toxicity (52). This score was better able to stratify risk of toxicity than Karnofsky PS alone. Both of these predictive tools are exciting because they provide information that can be used to discuss chemotherapy treatment with elderly patients. Unfortunately, neither has been validated outside of the initial population and, therefore, widespread adoption is not yet justified.

Subsequent trials have evaluated doublet chemotherapy regimens. Due to concerns about toxicity, these trials initially examined non-platinum chemotherapy combinations. The Southern Italy Cooperative Oncology Group (SICOG) conducted an RCT comparing gemcitabine and vinorelbine in combination to vinorelbine alone (56). The combination arm reported a median survival of 29 weeks compared to 18 weeks for vinorelbine alone. While this difference was statistically significant, there were concerns that the control group had worse survival than expected. These concerns prompted the multicenter Italian lung cancer in the elderly study (MILES), which compared three arms: vinorelbine plus gemcitabine, gemcitabine alone, and vinorelbine alone (57). Whereas, the SICOG trial enrolled 120 patients, MILES randomized 698 patients between the three arms. Median survivals were 30, 28, and 36 weeks for each of the arms, respectively. There was no statistically significant difference. There was, however, greater toxicity in the combination arm, specifically for neutropenia, thrombocytopenia, anemia, vomiting, constipation, and hepatic toxicity (57). These results do not support the use of the combination of vinorelbine and gemcitabine.

Further evaluation of doublet chemotherapy in patients ≥70 years was pursued. In one RCT of elderly patients or those with ECOG PS 2, the combination of gemcitabine and paclitaxel improved median survival to 9.2 months compared with 5.1 months for gemcitabine alone (58). However, a RCT in the same mixed population comparing gemcitabine/docetaxel to weekly docetaxel reported no difference in survival (59). A systematic review with meta-analysis of RCTs comparing non-platinum doublets with single-agent therapy for elderly patients showed no survival advantage to doublet therapy and higher risk of thrombocytopenia (60). Perhaps the most promising regimen was the combination of carboplatin with paclitaxel. One phase II trial demonstrated that weekly paclitaxel combined with carboplatin resulted in a 14-month median survival with quite manageable toxicity (61). Interestingly, when compared to standard paclitaxel, weekly paclitaxel appears to have equivalent benefit, but reduces the risk of neutropenia and peripheral neuropathy (62). The landmark trial investigating the use of platinum doublets in the elderly is Intergroupe Francophone Cancérologie Thoracique (IFCT)-0501 (63). This trial included 451 patients aged 70–89 years, with locally advanced or metastatic NSCLC and a PS of ECOG 0–2. Patients were randomized to carboplatin and weekly paclitaxel vs. monotherapy with either gemcitabine or vinorelbine. The trial was stopped early after interim analysis demonstrated superiority for the doublet regimen. Median overall survival was 10.3 months for carboplatin and paclitaxel compared to 6.2 months for monotherapy (hazard ratio 0.64, 95% CI 0.52–0.78, p < 0.0001). The largest increases in toxicity for doublet chemotherapy were neutropenia (48.4 vs. 12.4%) and asthenia (10.3 vs. 5.8%) (63). Point estimates were quite consistent for all subgroups and multivariable analysis confirmed expected prognostic factors like sex, PS, adenocarcinoma histology, and smoking history.

One final trial, conducted in Japan by Takeda et al., has only been published in abstract form (64). The trial enrolled 276 patients, who were chemotherapy naïve, age >70 years, ECOG PS 0–1, and with stage III/IV NSCLC. Patients were randomized to receive either docetaxel every 3 weeks or weekly cisplatin–docetaxel. Enrollment was stopped early due to futility, with median survival times of 13.3 months for cisplatin–docetaxel and 17.3 months for docetaxel alone (hazard ratio 1.557, 95% CI 0.976–2.485). Interestingly, neutropenia was far more common with docetaxel alone than the doublet regimen 88 vs. 11%. The survival of the monotherapy group was remarkably high compared previous trials (65–67).

The majority of subgroup analyses from earlier trials suggest that survival is similar, though not always equal, between younger and older patients with advanced NSCLC who receive the same chemotherapy (68–71). The evidence suggests that the results of IFCT-0501 should form the standard of care for first line chemotherapy treatment in fit elderly patients with advanced NSCLC, especially when no molecular abnormalities are detected.

In the second line setting, there are no elder-specific trials. A retrospective analysis of the JMEI trial comparing docetaxel to pemetrexed was completed for patients ≥70 years old vs. younger patients. Median survival of 9.5 and 7.7 months was reported for elderly patients receiving pemetrexed (n = 47) and docetaxel (n = 39). In younger patients, these values were 7.8 and 8.0 months. Febrile neutropenia occurred in only 2.5% of elderly patients receiving pemetrexed, but 19% of those being treated with docetaxel (p = 0.025) (72).

Bevacizumab is a monoclonal antibody that targets vascular endothelial growth factor (VEGF) and can be used in combination with first line platinum-based chemotherapy. Two trials, ECOG 4599 and AVAiL, originally tested the addition of bevacizumab to standard chemotherapy (73, 74). Analyses of the elderly patients in both of these trials were conducted. The AVAiL trial, comparing cisplatin and gemcitabine with or without bevacizumab, showed no improvement in overall survival in the elderly population (n = 304), with the addition of bevacizumab (73). An analysis of patients ≥70 years in the ECOG 4599 trial of carboplatin and paclitaxel with or without bevacizumab reported overall survival was 11.3 months with bevacizumab and 12.1 months without. There was a higher incidence of bleeding, neutropenia, and proteinuria in older compared to younger patients (74). There does not appear to be compelling evidence to include bevacizumab for those older than 65–70 years of age.

Molecularly defined subtypes of NSCLC have become incredibly important to management over the last 5 years. Epidermal growth factor receptor (EGFR) mutations are detected in approximately 15% of Caucasian patients with advanced NSCLC and these mutations are found more often, but not exclusively in younger, never smoking women, or those of Asian ethnicity (75). The presence of an EGFR mutation is highly predictive of benefit from EGFR tyrosine kinase inhibitors (TKIs) (76). More widespread screening of all NSCLC tumor samples for molecular abnormalities will increase the number of EGFR mutations identified in the elderly. Available data demonstrate EGFR TKIs (erlotinib, gefitinib, or afatinib) result in better progression free survival (PFS) and favorable toxicity compared to chemotherapy in NSCLC patients with an EGFR mutation (76–79). While few studies have examined the effect of this strategy exclusively in elderly patients, available data would suggest that elderly patients have similar response rate and PFS (80–84). Toxicities reported were the expected diarrhea, rash, and risk of transaminitis.

The NCIC BR.21 trial evaluated erlotinib in NSCLC patients who progressed after one or two prior chemotherapy treatments regardless of EGFR mutation status. The improvement in overall survival was seen in both EGFR mutated and wild type patients. A retrospective analysis of treatment effect and age in BR.21 found no statistically significant difference in treatment effect between younger and older patients for overall survival. Elderly patients did experience more Grade 3–4 toxicity (35 vs. 18%, p < 0.001). Based on this subgroup analysis, erlotinib seems to be a reasonable option for elderly patients in the second or third line setting (85). A trial in vulnerable elderly patients by CGA adds further support to this opinion, since both gemcitabine followed by erlotinib or the reverse on progression showed similar survival and tolerability (45).

The other actionable mutation found in NSCLC is a translocation in echinoderm microtubule associated protein-like 4 – anaplastic lymphoma kinase (EML4–ALK) gene, which is found in approximately 4% of patients with adenocarcinoma (75). Data in younger patients have been extremely promising with the use of crizotinib for EML4–ALK translocated NSCLC (86–88). Few patients were older than 70 years. A phase I study including 149 patients did report a response rate of 65% (40.8–84.6%) in patients ≥65 years (87). More recently, two other early-phase clinical trials with different ALK TKIs demonstrated response rates >50%, with ceritinib showing impressive responses even in crizotinib resistant disease (89, 90). While there are little data in elderly patients, there is no reason to believe this group would derive less benefit from ALK TKI therapy.