Second-line and Third-line Chemotherapy for Lung Cancer

Tag : pemetrexed disodium

Lung cancer is the second most common cancer in the United Statesand is the most common cause of cancer-related death. The AmericanCancer Society 1 estimated that there would be 213,380 new cases of lung cancerdiagnosed in the United States and 160,390 deaths from this diseasein 2007. Most of these individuals have regional or advanceddisease at diagnosis and are eligible for chemotherapy. Many whobegin chemotherapy show signs of disease progression or mustdiscontinue first-line therapy because of intolerable adverseeffects and are eligible for subsequent chemotherapy with differentagents. Second-line chemotherapy can produce symptomaticimprovement, objective responses, and modest improvements insurvival for patients with advanced disease. 2-5 Despite the availability of second and third-line therapies, thereis disagreement about the best treatment strategy.6 Little is knownregarding patterns of chemotherapy use and cost of care forpersons who receive such treatment.
Using data from a large nationwide health insurer, we sought todetermine the most common forms of second- and third-linechemotherapy use for persons with lung cancer. We also evaluatedkey services utilization (hospital days, ambulatory visits, andemergency department visits), and lifetime costs of care followinginitiation of the second-line chemotherapy regimen to determine if1 or more regimens had a significant effect on costs of carerelative to others (data not shown).
METHODS
Database and Patient Population
We identified patients with lung cancer from a health insurancedatabase (i3, Ingenix, UnitedHealthcare) representing privatelyinsured patients and individuals enrolled in Medicare + Choice andMedicaid across all 50 states, the District of Columbia, and the USVirgin Islands. 7 We searched administrative records for patients diagnosed ashaving lung cancer between January 1, 2002, and December 31, 2006,who had received at least 1 chemotherapy agent since diagnosis,using International Classification of Diseases, Ninth Revision,Clinical Modification (ICD-9-CM) codes for malignant neoplasm oftrachea, bronchus, and lung; malignant neoplasm of pleura; orcarcinoma in situ of respiratory system (codes 162.xx, 163.xx, and231.xx). To eliminate “false positives,” patients withICD-9-CM lung cancer codes for whom lung cancer was being ruled outor patients being treated for other cancers, we required claimsrecords with ICD-9-CM lung cancer codes on at least 5 separatedates of service and no more than 10 dates of service withdiagnoses of other common malignant neoplasms that are treated withchemotherapy recorded between 30 days before and 6 months after theinitial lung cancer diagnosis ( Figure 1 ).

First-line Chemotherapy Use
The ICD-9-CM, Current Procedural Terminology 4 , and Healthcare Common Procedure Coding System (HCPCS) level IIprocedure and diagnosis codes were utilized to identifychemotherapy administration cycles. Selecting cases for analysisproved to be problematic due to variation in coding methods amongservice providers, most notably relating to the common use ofICD-9-CM diagnosis code V58.1 (encounter for antineoplasticchemotherapy and immunotherapy), which has been previously noted tobe an unreliable indicator of chemotherapy administration. 8,9 Initially, patients were excluded if they had only chemotherapyadministration codes for a date of service with no agent specified.However, we found many cases in which dates of service withICD-9-CM code V58.1 and HCPCS agent codes were closely followed bydates of service for which V58.1 was the only code denoting thevisit as a chemotherapy encounter. Often this pattern was repeatedseveral times. Closer examination of the service dates without anagent often revealed laboratory, diagnostic, and palliativeprocedures related to chemotherapy management. Based on thisobservation, we surmised that providers used ICD-9-CM code V58.1for chemotherapy-related services and for chemotherapyadministration.
We allowed patients in our analysis set to have up to 4 consecutivedates of service at the beginning of their regimen in whichICD-9-CM code V58.1 was used to indicate a chemotherapy encounterbut no specific agents were recorded, provided that afterward atleast 3 service dates were coded with specific agents. Subsequently, patients with more than 8 consecutive nonspecificchemotherapy encounters were excluded from the analysis. Commonfirst-line agent combinations were based on those listed in theNational Cancer Institute’s Physician Data Query (PDQ)database. 10,11
Second-line and Third-line Chemotherapy
Second-line therapy is administered for disease progression,recurrence, or intolerable adverse effects following administrationof initial chemotherapy. We defined a second-line chemotherapyagent in 1 of 2 ways. First, a second-line agent could beinjectable; if so, then the agent was one with an initialinjectable administration date that was at least 28 days followingthe last delivery date of a different first-line chemotherapy agentor widely recognized combination of agents. Second, oral agentssuch as erlotinib hydrochloride and gefitinib, which are used afterfailure of first-line chemotherapy, were counted as second-lineagents if the initial date on which the prescription was filledoccurred on or after the last date of administration of first-linechemotherapy.
For inclusion, second-line regimens had to meet the same criteriaas the first line, as well as having no more than 3 uncodedchemotherapy visits between the last recorded date of thefirst-line treatment and the initiation of the second. If thepatient received chemotherapy beyond his or her first-linetreatment and the pattern of second-line administration did notmeet the criteria, the patient was excluded from the analysis. Welooked specifically for US Food and Drug Administration(FDA)-approved second-line therapies, including docetaxel,pemetrexed disodium, erlotinib, and gefitinib.
We defined third-line chemotherapy as any third chemotherapy agentwith an initial administration date following 2 separate courses oftreatment. Because of heterogeneity in prescribing and difficultyin identifying regimens as defined by the PDQ or as labeled by theFDA, agents identified as third-line therapy represent only thefirst drug given in the third-line setting.
Evaluating Resource Use and Cost of Care
Resource utilization was based on claims. Costs were based onhealth plan reimbursements for all goods and services covered bythe plan. Key services were grouped into inpatient days, outpatientvisits, chemotherapy, emergency department visits, laboratory,outpatient pharmacy, and skilled nursing facility, hospice, or homecare days. The costs of chemotherapy represent injectablechemotherapy agents; the costs of agent administration and relatedprocedures are accounted for in other categories. Oral oncologicagents are captured under outpatient pharmacy. Laboratory costsinclude the actual processing of the specimen. Using theKaplan-Meier sample average (KMSA) estimator method, we estimated1-year and 2-year total costs for patients with lung cancer. 12 The KMSA estimator of the total cost of care is given by thefollowing equation: E = SiPi * Ei, where i denotes month fromdiagnosis (0 denotes the diagnosis month); Pi, the probability ofbeing observed in month i; and Ei, the mean cost incurred in monthi among all cases observed to this time. Ei includes the costs ofcases surviving through month i and of cases dying in month i.Total expected costs and variances around expected costs areincluded, using an expression for variance that was developed forthe KMSA estimator. 13 In accordance with standards for economic evaluation, results forall cost estimates are presented with costs in future yearsdiscounted at annual rates of 3%. 14
Utilization and the aforementioned key services costs are reportedfor lung cancer cases from the date of initial chemotherapyadministration of the patient’s first-, second-, orthird-line regimen. Data were analyzed for the first and secondyears from the date of a patient’s first-line therapy andwere further stratified by whether he or she had only first-linetherapy, first- and second-line therapy, or therapy beyond asecond-line regimen. Analysis beginning with the initiation ofsecond- and third-line chemotherapy regimens had only a 1-year timehorizon. To determine how the choice of initial therapy affectscost of care, subjects were stratified by initial type of therapy.
Because the marketing of gefitinib was restricted midway throughour analysis period, we only include persons receiving thismedication in the summary statistics. We exclude gefitinibrecipients from the logistic regression and utilization analyses.
RESULTS
The initial search yielded 30,798 individuals older than 20 yearswith at least 1 diagnosis code for lung cancer enrolled betweenJanuary 1, 2002, and December 31, 2006. Applying exclusion criteriaruled out diagnoses of other cancers and reduced the sample to11,953 patients; of those, 6289 had at least 1 claim recordindicating that they had received chemotherapy. After applyingrestrictions for having a minimum number of agents specified alongwith an administration code as already outlined, 2523 patients wereincluded in the analysis. Eighty-one percent of these individualswere covered by commercial insurance, 17% through Medicarecontracts, and 2% through Medicaid contracts.
The mean interval between the last date of administration of thefirst-line agent and the initiation of the second-line treatmentwas 114 days for injectable agents. Almost half (46%) of thepatients received second-line therapy within 60 days of the lastdate of administration of first-line therapy. An additional 24%received second-line therapy within 120 days of first-line therapy.Among the remaining 30%, the period between first and secondtreatments ranged from 121 to 844 days.
Among patients who had second-line therapy, patients receivingdocetaxel had recorded claims for the greatest number of months(mean, 8.8 months) after the initiation of therapy, while thosereceiving erlotinib had the fewest number of claims months (mean,5.4 months). Because erlotinib did not become available untilDecember 2004, a higher relative percentage of patients begantherapy toward the end of the observation period, resulting in ashorter observation time. For second-line therapy, an additional385 agents or chemotherapy combinations were observed in the data;the most common regimens are given in Table 1 .

Chemotherapy Prescribing Patterns
Of 2523 persons included in the analysis, the most commonlyprescribed first-line regimen was carboplatin and paclitaxel.Carboplatin in combination with etoposide phosphate, gemcitabinehydrochloride, and docetaxel was also commonly prescribed ( Table 2 ).

Of patients receiving first-line chemotherapy, 25% receivedsecond-line therapy. Docetaxel was the most common second-lineagent, but there was much more variation for second-linechemotherapy agents and combinations than for first-linetreatments. Gefitinib was the second most commonly prescribedsecond-line agent, but the use of gefitinib as a proportion of allsecond-line therapies declined rapidly during the period ofobservation after findings from clinical studies suggested that itdid not improve survival and after the subsequent FDA labelingchange. In contrast, erlotinib prescriptions increasedsubstantially between 2004 (the year of its FDA approval) and 2005( Figure 2 ).

Only 10% of first-line patients received third-line chemotherapy.Gefitinib was most commonly prescribed overall, but over timeerlotinib replaced gefitinib as the most commonly prescribedthird-line agent after the labeling of gefitinib was modified inJune of 2005 to severely reduce its access ( Figure 2 ).
The logistic regression analysis showed that age, choice offirst-line regimen, time to initial treatment, and diagnosis yearwere significantly associated with receiving second-line treatment.Older persons were significantly less likely to receive second-linechemotherapy, but the odds ratio (OR) was only marginally differentfrom 1 (OR, 0.97; 95% confidence interval [CI], 0.96-0.98).Likewise, increasing time from initial diagnosis to treatment wasassociated with lower likelihood of receiving second-line therapy(OR, 0.83; 95% CI, 0.75-0.92). Using cisplatin plus etoposide asthe reference regimen, we found that persons receiving carboplatinplus etoposide (OR, 0.56; 95% CI, 0.39-0.79) and carboplatin pluspaclitaxel (OR, 0.73; 95% CI, 0.55-0.97) were significantly lesslikely to receive second-line chemotherapy. No other majorfirst-line chemotherapy regimen was significantly associated withthe likelihood of receiving second-line chemotherapy.
Utilization and Cost for Those Receiving Second-line and Third-lineChemotherapy
We compared overall costs for persons receiving only first-linechemotherapy vs additional chemotherapy regimens. Patientsreceiving first-line therapy only were observed for a mean of 8.1months vs 11.8 months among those receiving first- vs second-linetherapy and for a mean of 16 months among persons receivingthird-line therapy. In the first 12 months following theadministration of chemotherapy, the costs per patient per month(PPPM) were similar for the different regimens among personsreceiving only first-line therapy ( Table 3 ), ranging from approximately $3500 to $3900. The PPPM costs in thefirst 12 months for those receiving first- and second-line therapywere highest for those receiving carboplatin plus docetaxel asfirst-line therapy and were lowest for those receiving carboplatinplus etoposide. The PPPM costs were incrementally higher for thosewho received third-line vs second-line vs first-line chemotherapyregardless of initial first-line regimen, with PPPM costs rangingfrom $7200 to $9600 for those receiving 3 lines of chemotherapy.


Far fewer individuals were observed in the second 12 months fromthe date of initial chemotherapy administration, reflecting poorsurvival for persons with lung cancer (Table 3). For those who wereobserved in the second 12 months, the total and PPPM costs wereincrementally higher for those receiving second and third-linetreatments.
Starting from the point of initial second-line treatment, PPPMreimbursements were highest for persons receiving pemetrexed assecond-line therapy and were lowest for persons receivingerlotinib. These results are summarized in Table 4 .

DISCUSSION
We conducted an analysis of patterns of utilization and costs ofsecond- and third-line chemotherapy for patients with lung cancerfrom the perspective of a health insurer. We find that few patientsreceive second- and third-line therapy. In general, the lifetimecosts of care for patients who receive chemotherapy do not varysubstantially according to their initial regimen. Rather, whatmatters is whether they receive second- or third-line treatment.
Monthly costs of care were substantially higher for those receivingsecond-line treatment and were still higher for patients receivingthird-line chemotherapy compared with costs for those receivingonly first-line treatment. Expenditures for all services (such asinpatient stays, outpatient visits, and emergency departmentvisits) increased for these individuals. Therefore, the total costsmay reflect more intensive management at all levels. Observationtimes were significantly longer for persons who received second-and third-line treatments compared with those who received onlyfirst-line treatment.
Several factors were significantly associated with the likelihoodof receiving second-line therapy. As has been found in otherpopulation-based analyses of lung cancer therapy, 15,16 we find that older persons were somewhat less likely to receivesecond-line therapy. Increasing time between diagnosis andfirst-line therapy was associated with a lower likelihood ofreceiving second-line therapy. It is somewhat difficult tointerpret this finding without clinical information, but delays ininitiating first-line chemotherapy may be a marker of reducedphysician or patient preference for chemotherapy in general or ofclinical factors that delay the use of chemotherapy (eg, poorlycontrolled cardiovascular disease). Persons who receivedchemotherapy that included a platinum-containing agent (carboplatinor cisplatin) were less likely to receive second-line therapy thanthose who did not. Platinum-containing agents are recommended forfirst-line therapy 17 but are poorly tolerated. Therefore, those who receive them inthis setting may be less likely to seek second-line therapy.
During the study period, the use of the tyrosine kinase inhibitorsgefitinib and erlotinib grew substantially as agents for second-and third-line therapies, replacing a proportion of injectablechemotherapy agents. Erlotinib seems to have made substantial gainsin utilization as a second- and third-line treatment for patientswith lung cancer. Following FDA restrictions on the use ofgefitinib, this drug is now all but removed from general use inthis setting. It is likely that most patients who were candidatesfor gefitinib are now being treated with erlotinib. It is unclearwhether erlotinib is replacing other chemotherapy agents in orderof treatment preference. It is clear that erlotinib is not beingused substantially as a third-line agent for those who failsecond-line therapy.
A modest proportion of patients who began chemotherapy receivedthird-line chemotherapy. Most patients were given fewadministrations of third-line agents, and the intervals betweenadministrations were irregular. This is understandable becausesurvival is poor overall for patients with lung cancer, and it islikely that most patients who showed signs of progression aftersecond-line treatment died before having the opportunity to receivethird-line treatment or they or their physicians decided that thecosts (monetary and treatment-related morbidity) of furthertreatment were not justified by the expected survival difference.
Limitations
The primary limitation of this analysis is the lack of clinicalinformation. Although administrative claims are accurate inidentifying persons who received chemotherapy, 9,18 claims cannot distinguish small cell and non–small cell lungcancer and cannot identify dates of death. Approximately 80% ofincident lung cancers demonstrate non–small cell histologicfindings, and the remainder are primarily of small cell type.Chemotherapy regimens differ for small cell and non–smallcell lung cancer, although similar agents are commonly used forfirst-line treatment. 10,11
Although the observation times were substantially longer for thosereceiving second- and third-line therapies, it is impossible toconclude that treatment resulted in greater survival. Selectioneffects could explain the difference; that is, those receivingadditional treatments received them because they were in betterhealth in general compared with those who did not receivesecond-line therapy. We can say that persons receiving second- andthird-line regimens were treated much more intensively overall, astheir PPPM costs were substantially higher during the observationperiods.
Implications
Non–small cell lung cancer comprises approximately 80% of alllung cancer cases. During the period of this analysis, 4 agents(docetaxel, pemetrexed, gefitinib, and erlotinib) were approved foruse as second-line therapy in patients with refractorynon–small cell lung cancer. For small cell lung cancer thatprogresses after initial treatment, agents that have shown activityas second-line treatment include oral etoposide,etoposide-cisplatin, cyclophosphamide-doxorubicin–vincristinesulfate, lomustine-methotrexate, paclitaxel, and topotecanhydrochloride. 19 Our data suggest that a substantial fraction of chemotherapy usedin second-line settings is inconsistent with FDA labeling or is notsupported by clinical studies. This determination, combined withthe finding of wide variation in the use of agents, suggests a needfor quality improvement. Of note, persons who received chemotherapythat is not FDA approved had higher costs than those who usedapproved therapies in the second-line setting, although theabsolute difference (stated as PPPM cost) is modest overall.