A Survey of Non-Small Cell Lung Cancer Patients with Meningeal Carcinomatosis in Japan: Incidence and Medical Resource Consumption

Lung cancer was the second most common cancer in Japanese men males and the fourth most common in Japanese women females in 2005, and the leading cause of cancer-related deaths in men and women in 2009. Lung cancer is generally classified into two types based on histology: non-small cell lung cancer (NSCLC) and small cell lung cancer. The majority (85%) of new cases of lung cancer are NSCLC [1,2].


Incidence of meningeal carcinomatosis
The potential follow-up period for each patient was at least 1 year from diagnosis of NSCLC. Diagnosis of MC was proven by detection of malignant cells in cerebrospinal fluid in microscopic cytology or typical findings in magnetic resonance imaging (MRI). The starting date of the analysis was defined as the day of diagnosis of NSCLC and the event day was defined as the day of diagnosis of MC. Survival cases were defined based on the last confirmed day of survival and untraceable cases were defined using the last confirmed day of survival before follow-up ended. The cumulative incidence of MC was calculated taking into account the competing risk of death. The cumulative incidence of MC was calculated for all NSCLC patients and for subgroups of NSCLC patients with a diagnosis of stage IIIB or IV and histology of adenocarcinoma considering the treatment with an EGFR-TKI before diagnosis of MC.

Cost analysis
The patients with MC were followed-up until October 1, 2010. The evaluation of medical resource consumption adopted the payer perspective and included direct medical costs from the diagnosis of MC until the time of death. Indirect costs (e.g. lost income) and intangible costs (e.g. pain and suffering) were not assessed. Direct medical costs were analyzed using a medical service fee table and the national health  insurance drug price table (2010 edition). No discounting was used as the treatment duration in most cases was less than one year. A direct cost comparison was performed between patients treated with and without an EGFR-TKI after diagnosis of MC. To adjust for differences in time, resource utilization data are expressed as costs (yen) per patient per month.
Total costs were defined as all costs in the analysis and were classified into six items depending on the fee-for-service and therapeutic purpose: basic medical care and management, home care, and dietary medical expenses and hospitalization were defined as "Hospitalization and outpatient costs; costs of medication and injections as "drug costs"; costs of medical treatment and surgery as "treatment / surgery costs"; costs of blood test and imaging as "examination costs"; costs of radiotherapy as "radiologic treatment costs", and other expenses as "other costs". These six components were added to give the total costs. Drug costs included only the costs of pharmaceuticals used for treatment [12]. Examination costs included those for drug and contrast media used in an examination.

Patient characteristics
The characteristics of all the patients with NSCLC and those also diagnosed with MC are shown in Table 1. There were more male patients in both of these populations and adenocarcinoma was the most common histological type, followed by squamous cell carcinoma. Most patients diagnosed with MC were at an advanced NSCLC stage of IIIB or IV. The smoking rate was 73% in all patients, but only 54% in patients diagnosed with MC. EGFR mutations were found in 36% of patients diagnosed with MC, which is somewhat high, but the accuracy of this proportion is unclear because the test was not performed in 60% of the patients.
The characteristics of the patients at the time of diagnosis of MC are shown in Table 2. The median age of the 28 patients was 60 years old, with a range of 36 to 79 years old. Performance status (PS) was 0-1 for these patients at the time of diagnosis of NSCLC, but PS in 16 of these patients had deteriorated at the time of diagnosis of MC. Of the 28 patients, 24 had radiological evidence or a history of brain metastases;  8 had concurrent diagnoses of brain metastases and MC; and 16 had a history of brain metastases, of which 9 had received whole brain irradiation before diagnosis of MC. Fourteen patients had a history of treatment with an EGFR-TKI before diagnosis of MC. The median interval between the diagnoses of lung cancer and MC was 490 days, with a range of 22-1520 days.

Incidence of MC
The follow-up rates at 1-year and 2-year were 99.5% and 86.7%, respectively. The cumulative incidences of MC calculated taking account of the competing risk of death were estimated to be 2.4% and 6.0% at 1 and 2 years after diagnosis of NSCLC, respectively (Figure 1). In subgroup analyses, these respective incidences were 3.0% and 8.4% in 232 patients with adenocarcinoma, 3.44% and 7.92% in 254 patients with stage IIIB or IV NSCLC, 2.96% and 8.54% in 102 patients with stage IIIB or IV NSCLC treated with an EGFR-TKI, and 3.60% and 7.29% in 152 patients with stage IIIB or IV NSCLC who did not receive an EGFR-TKI ( Figure 2).

Cost analysis
Among the 28 patients with MC, economic data for nine were incomplete because of hospital transfer. Therefore, use of medical resources was evaluated for 19 cases, for which the characteristics, treatment history and treatment after diagnosis of MC are shown in Table 3. These patients were divided into those treated with (EGFR-TKI group) or without (non-EGFR-TKI group) an EGFR-TKI after diagnosis of MC. The median survival times after diagnosis of MC were 127 days in the EGFR-TKI group and 40 days in the non-EGFR-TKI group. The EGFR-TKI group was younger and had a better PS compared with the non-EGFR-TKI group.
The median and range of total costs per patient per month were 706,607 (¥1,371,110-¥246,601) yen in the EGFR-TKI group and 1,033,361 (¥1,143,610-¥680,244) yen in the non-EGFR-TKI group (Table 4). Especially, there was a strong relationship between hospitalization costs and total costs. Drug costs were more than twofold higher in the EGFR-TKI group compared to the non-EGFR-TKI group. Excluding "Hospitalization and outpatient costs", the median and range costs per month per person were ¥361,519 (¥679,097-¥212,826) yen in the EGFR-TKI group and ¥287,733 (¥384,651-¥188,598) yen in the non-EGFR-TKI group. Management of increased intracranial pressure and pain control were performed for most cases. There was no case of drug-induced interstitial pneumonia. The median cost for infection was higher in the EGFR-TKI group, whereas the median cost of the treatment for nausea and vomiting was higher in the non-EGFR-TKI group.

Discussion
MC develops in 5-8% of solid cancer cases, with most of these cases having a histologic type of adenocarcinoma. In the current study in 376 patients diagnosed with NSCLC, the histologic type was adenocarcinoma in 26 of 28 patients who developed MC and the risk of MC was similar to that previously reported [13,14]. The cumulative incidence of MC in all NSCLC patients was calculated taking into account death as a competing risk, but the value may have been underestimated because there were some patients in whom test for confirmation could not be performed due to severe systemic conditions, despite clinical findings indicating a strong suspicion of MC, and patients diagnosed based on CT alone were excluded. Our findings also suggested that treatment of NSCLC with an EGFR-TKI is a risk for development of MC. However, many patients who developed MC had also brain metastasis, which might be a risk factor for MC. Many recent reports have suggested the efficacy of an EGFR-TKI for brain metastasis and these drugs may have been used selectively for patients with brain metastasis. Therefore, further investigation of the association between use of EGFR-TKI and development of MC is necessary.
Regarding resource consumption after development of MC, the mean total cost per patient per month was lower for EGFR-TKI-treated patients. The cost for anticancer drugs accounted for more than half of the drug costs in the EGFR-TKI group. In Japan, it will be difficult to use biosimilar of EGFR-TKI, this trend of cost would be continuing in near future [15]. Glycerol and steroids were mainly used for the control of intracranial pressure control and accounted for the second largest cost following the cost of transfusion in the non-EGFR-TKI group. The cost for pain control differed markedly between the groups, suggesting that this cost is strongly dependent on the patient condition. Measures against constipation were taken in many patients, but these did not have a marked influence on costs. Skin eruption is a frequent adverse event in EGFR-TKI-treated patients and was expected to result in a marked difference in costs for skin control, but there was no major difference in these costs between the two groups. The costs for antiemetic treatment and sedation were higher in the non-EGFR-TKI group. Since many patients with relatively poor conditions were included in the non-EGFR-TKI group, the incidence of severe neurological symptoms may also have been high. Regarding infection, the incidence of infectious disease complication was high in the EGFR-TKI group and the cost for control against infection was also high. Therefore, these costs for infection and for anticancer drugs should be considered in calculating the costs for treatment of MC.
In the 28 MC patients, the median survival times were 164 days (8-16 weeks) for those treated with an EGFR-TKI and 40 days (4-6 weeks) for those who did not receive an EGFR-TKI. Therefore, the prognosis of MC was very poor [16], but the survival time was longer for EGFR-TKI-treated patients, as also found in previous studies, and 3 of these patients survived for longer than one year after diagnosis. However, PS and the presence or absence of serious neurological symptoms and metastasis in other organs are strong prognostic factors [17]. PS was 0-1 in many EGFR-TKI-treated patients and it is possible that EGFR-TKI administration was biased to patients in a relatively good condition for whom long-term survival could be expected.
Several limitations of the study warrant mention. First, this is an observational study which cannot exclude potential of confounding bias, so prescription could not be controlled and direct comparison between EGFR-TKI and non-EGFR-TKI groups may be difficult. Moreover, the patients were treated at a single institution and the sample size was small, suggesting the presence of a selection bias in analyzed cases. For evaluation of resource consumption, the 2010 payments for medical services and standard prices for medicines were used to calculate the cost. The drug price is reduced in revision in every 2 years, leading to potential of over estimation of costs. However, recent revisions do not have much impact. The analysis was also performed from the payer's viewpoint and only direct medical expenditure was investigated. It may also be necessary to evaluate the cost from a social point of view with inclusion of direct non-medical expenditures and associated costs for a relatively serious disease such as MC, including transportation expenses and time-costs for unpaid caregivers. Another concern is that EGFR testing was not performed in 60% of the patients, which is substantially different from the current clinical practice.