Short-term outcomes of craniotomy for malignant brain tumors in the elderly
Disparity in resection rates for malignant brain tumors in elderly patients is partially attributed to a belief that advanced age is associated with an increased risk of postoperative morbidity and mortality. The objective of this study was to investigate the effect of advanced age (≥75 years) on 30-day outcomes in patients with primary and metastatic brain tumors who underwent craniotomy for definitive resection of a malignant brain tumor.
The authors conducted prospective analyses of the American College of Surgeons' National Surgical Quality-Improvement Project (NSQIP) database from 2006 to 2010 of 970 patients aged ≥40 years who underwent craniotomy for definitive resection of neoplasm. Preoperative and intraoperative characteristics and 30-day outcomes were stratified by age. By using propensity scores, 134 patients (aged ≥75 years) were matched to 134 patients ages 40 to 74 years. Logistic regression was used to predict adverse postoperative outcomes.
The median length of hospital stay was 5 days; the rate of minor and major complications were 5.9% and 13.1%, respectively; 5.7% of patients returned to the operating room; and 4.3% of patients died within 30 days. Advanced age did not increase the odds for poorer short-term outcomes.
Advanced age did not increase the risk of poor outcomes after surgical resection of primary or metastatic intracranial tumors when analyses were controlled for other risk factors. These results suggest that age should not be used, in isolation, as an a priori factor to discourage pursuing craniotomy. Cancer 2013. © 2012 American Cancer Society.
Each year in the United States, greater than 20,000 patients are diagnosed with a primary malignant glial neoplasm, and between 100,000 and 200,000 patients are diagnosed with a new brain metastasis.1 Continued trends toward an aging population, combined with the increased incidence in tumors of all kinds with advancing age, lends greater importance to identifying the effect of advanced age on a patient's ability to withstand aggressive cancer treatments, including craniotomy for tumor.
Several studies have demonstrated that surgical tumor resection, with and without additional treatment, prolongs the survival of patients with both primary2,3 and metastatic brain tumors.4,5 However, a recent study conducted in a nationally representative sample of US adults6 confirmed a persistent disparity in surgical resection rates for the elderly. This inequity has been attributed in part to a belief that elderly patients are a more fragile population, unable to tolerate long surgeries, and have a higher risk of morbidity and mortality after craniotomy for malignant tumor resection.6–9 The objective of this study was to investigate the effect of advanced age (≥75 years), alone and in combination with several preoperative and intraoperative factors, on 30-day outcomes in patients with primary and metastatic brain tumors who underwent craniotomy for the definitive resection of a neoplasm.
MATERIALS AND METHODS
The medical records of all patients who underwent craniotomy for definitive resection of an intracranial primary or metastatic brain tumor that were included in the American College of Surgeon (ACS) National Surgical Quality Improvement Program (NSQIP) database between 2006 and 2010 were evaluated. A detailed description of the ACS-NSQIP database, including design, sampling strategy, and variable definitions, are available elsewhere.10 This study was approved by the Cleveland Clinic Institutional Review Board.
The study population consisted of 970 adult patients aged ≥40 years who underwent craniotomy between 2006 and 2010. We used Current Procedure Terminology codes to identify craniotomy patients who underwent definitive tumor resection and matched then with patients who had been assigned an International Classification of Disease-9th edition code that corresponded to malignant neoplasm. Of these 970 patients, 563 (58%) had a malignant glioma, 380 (39%) had a metastatic brain tumor, and 27 (3%) had a malignant tumor of a cranial nerve, cranial meninges, or not otherwise specified.
Age at surgery was categorized as ages 40 to 74 years and ≥75 years. Sensitivity analyses were conducted using age as categorized by decade and altering the definition of elderly to ≥60 years and ≥65 years.
Patient History Covariates
We assessed all available preoperative and intraoperative factors that were identified previously as having an effect on postoperative outcomes.11 Transfer status was dichotomized as admitted from home versus transferred from another facility (including an acute care hospital, a chronic care facility, or an outside emergency department). Alcohol intake was recorded as having consumed ≥2 drinks per day in the 2 weeks before admission. Functional status captured the ability to perform activities of daily living within 30 days before surgery and was dichotomized as independent versus partially or totally dependent. American Society of Anesthesiologists physical status classification values were dichotomized as “1 + 2,” indicating a normal, healthy patient or a patient with mild disease, and “3 + 4,” indicating a patient with severe, systemic disease that was or was not life threatening.
Patients who required ventilator-assisted respiration during the 48 hours before surgery, had been diagnosed with chronic obstructive pulmonary disease, and/or had evidence of pneumonia at the time of surgery were considered to have pulmonary comorbidities. Cardiovascular comorbidities were considered positive for patients who had the following: a diagnosis of congestive heart failure within 30 days before surgery, myocardial infarction in the 6 months before surgery, percutaneous coronary intervention, previous cardiac surgery, self-reported angina in the month before surgery, an angioplasty or revascularization procedure for atherosclerotic peripheral vascular disease, and/or an experience of rest pain or gangrene. Dyspnea was self-reported as difficult, painful, or labored breathing with moderate exertion. Patients who had acute or chronic renal failure requiring treatment with peritoneal dialysis, hemodialysis, hemofiltration, hemodiafiltration, or ultrafiltration within 2 weeks before surgery were considered to have renal comorbidities. Central nervous system comorbidities were reported positive for patients who had coma for ≥24 hours, hemiplegia, transient ischemic attacks, cerebrovascular accident, paraplegia, or quadriplegia.
Chemotherapy and radiotherapy for cancer within 90 days before surgery were individually captured. Self-reported patient history of abnormal bleeding, self-reported family history of bleeding disorders, vitamin K deficiency, and a comprehensive list of medications that pose a risk for bleeding abnormalities were captured through the NSQIP variable “bleeding disorders.” Preoperative hemostatic screening laboratory values were recorded in the NSQIP database if drawn within 90 days before the surgical procedure. Intraoperative and postoperative transfusions were aggregated, using an erythrocyte amount received of ≥250 mL as indicative of transfusion.
Outcomes of Interest
The time between hospital admission and surgery, operation and discharge, and total length of stay (LOS) were individually assessed. Prolonged LOS was defined as LOS >75% of the patients in the sample, which was 9 days. Postoperative complications were defined as those that occurred within 30 days of surgery. Minor complications were 1 or more of the following: superficial surgical site infection, urinary tract infection, deep venous thrombosis, or thrombophlebitis. Major complications were 1 or more of the following: deep incision surgical site infection, organ or space surgical site infection, wound disruption, pneumonia, unplanned intubation, pulmonary embolism, >48 hours of postoperative ventilator-assisted respiration, progressive renal insufficiency, acute renal failure, cardiovascular accident with neurologic deficit, coma for >24 hours, peripheral nerve injury, cardiac arrest requiring cardiopulmonary resuscitation, myocardial infarction, graft, prosthesis or flap failure, sepsis, septic shock, and/or 30-day return to the operating room (OR). The total number of postoperative complications per patient also was reported. Return to the OR (yes/no) was defined as any unplanned return to the OR for a surgical procedure within 30 days postoperatively. “Days of operation until death” was used to dichotomize mortality at 30 days postoperatively.
Preoperative and intraoperative outcomes and 30-day outcomes were compared across age groups using Pearson chi-square tests for categorical variables; analysis of variance was used for continuous variables. Propensity scores, including all variables listed in Table 1, were generated to obtain an approximately unbiased measure of the effect of age on adverse outcomes in the group ages 40 to74 years versus the group aged ≥75 years. A 1:1 greedy matching technique12 was used to match all elderly patients (aged ≥75 years) with a unique patient between ages 40 and 74 years. Covariates were compared between age groups in the matched sample. Logistic regression analysis was used to test whether age was independently associated with adverse outcomes. Covariates that remained unbalanced after matching on propensity scores were included in the final models. In addition, all analyses were repeated separately for patients with primary glioma and patients with metastatic brain tumors. P values < .05 were considered statistically significant. The SAS statistical software package (version 9.2; SAS Institute Inc., Cary, NC) was used for all statistical analyses.
Table 1. P Values for Covariate Balance Between Age Groups Before and After Stratification on Propensity Score
|Variablea||Ages 40–74 Years, N = 836||Aged ≥75 Years, N = 134||Unstratifiedb||Stratifiedc|
|Percentage of overall cohort||86.2||13.8|| || |
|Admitted from home||90.2||85.1||.07||.62|
|Smoking status|| || ||< .001d||.76|
|Current||27.5||9.7|| || |
|Previous||26.9||39.6|| || |
|Never||45.6||50.8|| || |
|Independent functional status||80.7||67.9||< .001d||.32|
|ASA classification|| || ||< .001d||1.0|
|1+2||19.6||3.7|| || |
|3+4||79.7||96.3|| || |
|BMI: Mean±SD, kg/m2||28.2 ± 6.3||26.8 ± 5.6||.02d||< .001d|
|Pulmonary comorbidities||8.5||19.4||< .001d||.64|
|Cardiovascular comorbidities||7.5||23.9||< .001d||.56|
|Hypertension requiring medication||42.2||76.1||< .001d||.57|
|Diabetes mellitus||11.4||23.1||< .001d||1.0|
|Weight loss >10% in 6 mo before surgery||3.7||7.5||.04d||.82|
|Steroid use for chronic condition||26.3||20.2||.13||.65|
|Preoperative chemotherapy and/or radiotherapy||10.1||1.5||.001d||1.0|
|Prior operation within 30 d||2.4||1.5||.80||.90|
|Abnormal albumin||11.2||24.6||< .001d||.75|
|Abnormal alkaline phosphatase||5.3||4.5||.91||.88|
|Abnormal creatinine||3.7||14.9||< .001d||.73|
|Abnormal platelet count||11.2||7.5||.35||.81|
|Abnormal total bilirubin||3.2||3.7||.67||.57|
|Abnormal white blood cell count||39.1||43.3||.55||.76|
|Level of residency supervision in the OR|| || ||.23||.86|
|Attending alone||48.7||55.2|| || |
|Attending and resident||50.7||43.3|| || |
|Wound class|| || ||.48||1.0|
|Clean||96.5||98.5|| || |
|Clean-contaminated||1.7||0.8|| || |
|Contaminated or infected||1.8||0.8|| || |
Of the 970 patients in the current analysis who underwent craniotomy for definitive resection of a malignant, intracranial brain tumor, 134 patients (13.8%) were aged ≥75 years. Compared with patients ages 40 to 74 years, elderly patients (aged ≥75 years) were slightly more likely to be men and were less likely to be admitted directly from home, to be current smokers, and to have independent functional status or an American Society of Anesthesiologists classification of 1 or 2 (Table 1). Patients aged ≥75 years had slightly lower mean baseline body mass index (BMI) compared with patients ages 40 to 74 years (26.8 kg/m2 vs 28.2 kg/m2). The elderly had higher rates of pulmonary, cardiovascular, and central nervous system comorbidities; hypertension requiring medication; and diabetes mellitus. They also were less likely to have received preoperative chemotherapy or radiation (1.5% of elderly compared with 10.1% in the group ages 40–74 years). The frequency of emergency cases was similar between age groups. Elderly patients had a higher prevalence of abnormal preoperative laboratory values, including sodium, creatinine, and albumin levels (Table 1). There were no significant differences in preoperative alkaline phosphatase levels, bilirubin levels, hematocrit, white blood cell counts, or platelet counts across age groups nor between the frequency of emergency procedures, the level of residency supervision during surgery, or wound classification. The prevalence of intraoperative factors consisting of the level of residency supervision, wound class, and transfusions undergone, was similar between age groups.
The 30-day outcomes for the entire group were 5.9% (n = 57) of all patients who experienced 1 or more minor complication, 13.1% (n = 127) of those who experienced 1 or more major complication, 5.7% (n = 55) of those who returned to the OR within 30 days, for an overall 4.3% (n = 42) 30-day mortality rate (Table 2). Although the total length of hospitalization did not differ significantly between age groups, elderly patients were more likely to have a prolonged LOS (41%) (Table 2). No association was observed between age and minor, major, or total number of complications; and no difference was observed in the percentage of elderly patients returning to the OR or dying within 30 days postoperatively.
Table 2. Thirty-Day Postoperative Outcomes Stratified by Age Groups, N = 970
|Variable||Ages 40–74 Years, N = 836||Aged ≥75 Years, N = 134||Pa|
|Percentage of overall cohort||86.2||13.8|| |
|Time between hospital admission and operation: Median (IQR), d||0 (0–3)||2 (0–4)||.76|
|Time between operation and discharge: Median (IQR), d||3 (2–6)||5 (3–7)||.28|
|Total length of hospital stay: Median (IQR), d||5 (3–8)||7 (4–12)||.41|
|Prolonged length of hospital stay >9 d||24.2||41||< .001b|
|Total no. of complications|| || ||.09|
|Return to the OR||5.3||8.2||.17|
All covariates listed in Table 1 were used to generate the propensity score, and elderly patients (aged ≥75 years) were matched 1:1 with patients ages 40 to 74 years. After matching, BMI remained the only covariate that had significantly different values between age groups. The relation between age and each adverse outcome of interest was determined using logistic regression at baseline, after matching by propensity scores, and further including BMI (Table 3). Although, in the unmatched analyses, elderly patients had an odds of 2.3 (95% confidence interval, 1.5–3.2) for prolonged LOS compared with patients ages 40 to 74 years, this difference became nonsignificant after propensity matching. There was no relation between age and any of the adverse outcomes in the propensity-matched sample with and without BMI included (Table 3).
Table 3. Age Group Comparisons Between Patients Aged 75 Years Versus Those Ages 40 to 75 Years for Adverse Outcomes (Odds Ratios) Using Different Analysis Methods
|Adverse Outcome||Logistic Regression Analysesa||Propensity-Matched Sample||Propensity-Matched Sample Plus BMI|
|Prolonged LOS: >9 d||2.2 (1.5–3.2)b||1.3 (0.8–2.2)||1.1 (0.7–1.9)|
|Minor complications||1.0 (0.5–2.2)||0.5 (0.2–1.3)||0.5 (0.2–1.4)|
|Major complications||1.3 (0.8–2.1)||1.0 (0.5–1.8)||0.9 (0.5–1.8)|
|Total no. of complications||1.2 (0.8–1.9)||0.8 (0.5–1.5)||0.8 (0.4–1.5)|
|30-D return to the OR||1.6 (0.8–3.2)||1.2 (0.5–3.1)||1.2 (0.4–3.0)|
|30-D mortality||1.8 (0.8–3.7)||1.3 (0.5–3.6)||1.0 (0.4–2.9)|
Sensitivity analyses with the sample limited to those with malignant glioma (n = 563) and those with bone metastases (n = 380) individually produced consistent results. We also used age categorized by decade and altered the definition of elderly to ≥60 years and ≥65 years. By using different modeling techniques, including multivariate logistic regression analyses and backward selection model building, we produced the same findings. Extensive analyses of potential interactions between age and other preoperative and intraoperative variables revealed no significant effects.
In this study, advanced age (≥75 years) was not associated with poorer operative or 30-day outcomes in patients who underwent craniotomy for definitive resection of a malignant intracranial tumor. Patients aged ≥75 years fared equally as well as patients ages 40 to 74 years, which is the typical age range for most adults with a malignant brain tumor who undergo aggressive therapy and the population that is usually included in most, if not all, clinical trials.
Interpretations in the Context of the Literature
Elderly patients frequently are excluded from clinical trials,13–15 such as the 2005 trial by Stupp et al,16 who investigated the addition of temozolomide to radiotherapy in patients with glioblastoma. Although a few trials since then have included older patients,17,18 the focus those trials has been on the application of adjuvant therapy after nondefinitive, attempted or completed gross total resection of tumor. It has been demonstrated that gross total resection has a survival advantage compared with partial resection and certainly compared with biopsy alone.2,3
Few studies have explored the association between age and short-term postoperative outcomes, and the reports have been conflicting.19–22 By using samples from patients who underwent resection of both primary and metastatic tumors, 2 studies associated age with an increased risk of major complications.7,9 Two other studies, which included only patients who had bone metastases, reported an association between age and increased morbidity and mortality.19,22 In contrast, 2 studies in patients with malignant glioma concluded that advanced age was not associated with increased morbidity, and no association was observed between age and postsurgical neurologic function.20,21 Major limitations of previous study designs included limited assessment of preoperative and intraoperative factors, single-institution patient samples,7,9,21,22 small sample sizes with few events,7,21,22 and reliance on univariate analyses to arrive at the conclusions.7,13,21,22
We identified a greater prevalence for prolonged LOS among elderly patients across our entire sample; however, there was no significant difference between age groups when we examined total hospital LOS as a continuous variable (Table 2). No association between age and prolonged LOS remained after matching by propensity scores. Although elderly patients did have somewhat longer lengths of hospitalization with a mean of 2 days longer for patients aged ≥75 years compared with those ages 40 to 74 years, this difference was nonsignificant and disappeared completely after adjusting for other preoperative and intraoperative variables.
Although several studies have demonstrated enhanced survival in patients with both primary2,3 and metastatic brain tumors,4,5 there remains controversy about the risk for surgical complications in the elderly, causing inequities in surgery rates. Our study suggests that the perception that elderly patients have a higher risk for morbidity and mortality after craniotomy6–9 is unsupported. To our knowledge, only 1 previous study has reported an association between age (>60 years) and increased odds for major regional complications9 in craniotomy patients. This finding was published almost 15 years ago and came from a single institution with a limited sample size, a small event rate, and a limited number of preoperative and intraoperative variables analyzed. Our study revisits this topic and indicates that there is no effect of age on 30-day morbidity and mortality in patients who undergo craniotomy for resection of primary or metastatic tumor after controlling for other risk factors, and our results caution against the use of age as an a priori factor to discourage pursuing craniotomy.
The NSQIP database contains only patients who underwent surgery; therefore, we were not able to capture any patient who did not undergo surgery because of pre-existing risk factors, including advanced age, as has been reported in other surgical specialties.23 We cannot account for issues of surgeon and hospital volume, which are known to influence postoperative morbidity and mortality after craniotomy for brain tumor,24 because it is ACS policy to maintain confidentiality for data-reporting institutions.10 However, we did model for the presence or absence of resident participation in the OR, which is a surrogate for academic versus nonacademic hospitals and tends to correlate with hospital size and volume, and which was not identified as different between age groups (Table 1) nor related to any of our outcomes of interest (data not shown). In addition, the low levels of postoperative complications identified in our sample were similar to those at high-volume facilities.24 Third, NSQIP only provides the last set of laboratory results before surgery, within 90 days of surgery; and it does not include certain features (such as the presence of a neurologic deficit, rather than overall “functional status”) that may influence outcome, disposition, and survival. Fourth, we did not have data on the size, exact location, number, or type of primary lesions for patients who had metastasis (n = 380); whether patients received whole-brain radiotherapy before resection; or the extent of resection, all of which have been previously identified as affecting survival.2–5 However, postoperative complications and short-term mortality reportedly were similar between patients who underwent craniotomy for a single tumor resection versus multiple tumor resections4and, although it has been observed that the type of primary lesion affects overall survival,25 no study to date has demonstrated a difference in postoperative short-term outcomes. Fifth, because there are suboptimal data on preoperative and postoperative neurologic function, these results cannot be compared against studies that only assessed neurologic function as a postoperative complication.20,21 Finally, the surgical population captured by NSQIP may not be wholly representative of the US neurosurgical population, because self-selected institutions contribute patient data. However, the sex and race distributions in the NSQIP database are representative of the US population, and data are collected prospectively from a substantial number of various types of institutions, thus providing a large and diverse sample size.10 Additional benefits of the NSQIP database are that data are collected in a standardized manner at each site with strict variable definitions and annual quality checks10and the database has been validated for accuracy and reproducibility and achieves a >95% 30-day outcome follow-up rate across consecutive cycles.10,26
In conclusion, to our knowledge, this is the first prospective, multi-institutional study to assess the relation between age, other preoperative and perioperative risk factors, and short-term outcomes after craniotomy for definitive tumor resection in the elderly (aged ≥75 years) compared with younger adults. Advanced age (≥75 years) was not correlated with poorer short-term outcomes. Although a few studies have assessed the role of age on the short-term outcome of patients with a primary or metastatic tumors, these had numerous limitations.7,9,19–22 Because patients aged 75 years (and, in some cases, aged 65 years or ≥70 years) usually or frequently are excluded from most clinical trials,13–15 there is a paucity of data regarding the risk of definitive surgical resection and its correlation with short-term, perioperative outcomes in these patients. Contrary to common assumptions, our analysis of a large, prospective, multi-institutional database suggests that advanced age does not predispose individuals undergoing aggressive surgical therapy for primary or metastatic intracranial tumor to increased risk for operative or short-term postoperative morbidity or mortality.
Dr. Seicean received support from an Agency for Healthcare Research and Quality (AHRQ) institutional training grant (T32-HS00059-14). Dr. Weil received support from the US Department of Defense Breast Cancer Research Program (grant W81XWH-062-0033) and from the Melvin Burkhardt chair in neurosurgical oncology and the Karen Colina Wilson research endowment within the Brain Tumor and Neuro-oncology Center at the Cleveland Clinic Foundation.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.