Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in older adults

JGO-00856; No. of pages: 7; 4C: Journal of Geriatric Oncology xxx (2019) xxx

Contents lists available at ScienceDirect

Journal of Geriatric Oncology

Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in older adults Redi Rahmani a,1, Samuel B. Tomlinson b,1, Gabrielle Santangelo a, Kwanza T. Warren b, Tyler Schmidt a, Kevin A. Walter a, G. Edward Vates a,⁎ a b

Department of Neurosurgery, University of Rochester Medical Center, Rochester, NY, United States of America School of Medicine and Dentistry, University of Rochester Medical Center, Rochester, NY, United States of America

a r t i c l e

i n f o

Article history: Received 29 March 2019 Received in revised form 6 July 2019 Accepted 25 October 2019 Available online xxxx Keywords: Malignant glioma Mortality Glioblastoma multiforme Frailty ACS-NSQIP

a b s t r a c t Introduction: Craniotomy for tumor resection improves survival in adults aged ≥65 years with malignant glioma. However, the decision to attempt resection must be weighed against the near-term risks of surgery. This study examined risk factors associated with unfavorable 30-day outcomes following craniotomy for malignant glioma resection in older adult patients. Materials and Methods: The American College of Surgeons National Surgical Quality Improvement Program database from 2012 to 2016 was queried for patients aged 65–89 years undergoing craniotomy for primary, supratentorial, malignant, intra-axial tumor resection. 30-day outcomes included mortality, life-threatening complication, unplanned readmission, reoperation, and change in living disposition. Independent risk factors were identified through multiple logistic regression. Results: In total, 1016 cases met eligibility criteria. Death occurred in 35 cases (3.4%). 58 patients (5.7%) suffered at least one life-threatening complication. Risk factors for morbidity and mortality included frontal lobe tumor, corticosteroid use, dependent functional status, and underweight body mass index (BMI). Among 816 patients admitted from home, 33.9% experienced a change in living disposition, which was associated with advanced age, female sex, frontal lobe tumor, underweight BMI, and diabetes mellitus (among others). Readmission (11.8%) was most frequently attributed to altered mental status, seizure, or venous thromboembolism. Reoperation was rare (4.5%). Discussion: Death and life-threatening morbidity were rare early outcomes for older adult patients undergoing malignant glioma resection. However, one in three patients admitted from home experienced a change in living disposition. Factors related to baseline state of health, tumor location, and corticosteroid regimen should be considered when anticipating the immediate postoperative course. © 2019 Published by Elsevier Ltd.

1. Introduction Malignant glioma, including glioblastoma multiforme (GBM), is the most common primary brain tumor in adults [1], and about half of GBM cases are diagnosed in patients aged 65 years or older [2,3]. The incidence of GBM in the older population has been increasing for decades [4,5], and advanced age at diagnosis is a prognostic factor associated with decreased survival [2]. As the population ages, malignant glioma is expected to account for an increasingly significant portion of the neurologic disease burden affecting older adult patients [3,6].

⁎ Corresponding author at: Department of Neurosurgery, University of Rochester Medical Center, 601 Elmwood Ave, Box 670, Rochester, NY 14642, United States of America. E-mail address: [email protected] (G.E. Vates). 1 These authors contributed equally to the work as co-first authors

Treatment standards for older adults with malignant glioma are open to debate [2,7]. Multiple studies indicate that tumor resection improves survival in older adults relative to biopsy or non-surgical interventions [8–13], though median survival is short (btwelve months) even with aggressive surgery [18]. While evidence indicates that older adult patients can withstand the short-term stress associated with glioma resection [14], other studies suggest that resection may be prohibitively morbid in older patients with diminished homeostatic reserves [15–18]. The likelihood of undergoing craniotomy for tumor resection decreases with age [19], which may partially reflect surgeons' reluctance to operate on older patients. Further, patients may prioritize near-time quality of life over the potential survival benefit associated with resection. Improved ability to anticipate the short-term postoperative course would help surgeons guide patients and families through these difficult treatment conversations. This study documented rates of adverse 30-day outcomes following craniotomy for resection of primary, malignant, supratentorial glioma in

https://doi.org/10.1016/j.jgo.2019.10.019 1879-4068/© 2019 Published by Elsevier Ltd.

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019

2

R. Rahmani et al. / Journal of Geriatric Oncology xxx (2019) xxx

older adult patients. Further, we elucidated risk factors associated with mortality, morbidity, and change in living disposition. Our sample comprised over 1000 cases documented in the American College of Surgeons National Surgical Quality Improvement Program (ACSNSQIP) database, a large national database capturing standardized peri-operative variables from hundreds of participating institutions. 2. Materials and Methods 2.1. Data Source The ACS-NSQIP database from 2012 to 2016 was the data source for this study. Preoperative variables related to patient demographics, medical co-morbidities, and routine laboratory data are submitted to this database by trained Surgical Clinical Reviewers (SCRs) from N700 institutions. 30-day outcome variables are tracked regardless of discharge disposition. Information about ACS-NSQIP auditing procedures are available online: http://www.facs.org/quality-programs/acs-nsqip/ program-specifics/data. This study was determined to be exempt from the University of Rochester Research Subjects Review Board. 2.2. Case Selection Current Procedural Terminology (CPT) code 61510 identified cases of craniotomy for excision of supratentorial brain tumor, excluding meningioma. Post-operative diagnosis of primary malignant brain neoplasm was determined using codes from the International Classification of Diseases (ICD), 9th and 10th revisions: ICD-9 (191.x) and ICD-10 (C71.x). Complete descriptions of CPT and ICD codes are available in Supplemental Table S1. The following inclusion criteria were enforced: (1) Age 65–89 years old; (2) neurosurgeon attending; (3) Nonemergency case designation; (4) General anesthesia; (5) Inpatient care; (6) Absence of known metastatic disease; (7) Operative time ≥90 min (in an attempt to exclude biopsy-only cases); (8) Labs drawn within 30 days of surgery; and (9) No missing data points of interest (see ‘Preoperative Characteristics’ below). This process yielded a cohort comprising 1016 cases of craniotomy for excision of supratentorial, malignant, intra-axial tumor. Gliomas account for ~80% of primary malignant brain tumors in adults [1]. Consistent with previous studies employing similar case selection criteria [20], we interpret our cohort as predominately composed of patients undergoing craniotomy for malignant glioma resection. 2.3. Preoperative Characteristics The ACS-NSQIP database tracks dozens of demographic and preoperative medical co-morbidities. Variables were selected based on (1) a priori suspicion of association with adverse outcomes; (2) uninterrupted documentation across the five-year sample window; and (3) representation in at least 1% of the overall sample. Selected variables included (with abbreviated description): age, sex, race/ethnicity, body mass index (BMI; kg/m2), diabetes mellitus (DM) on oral agent or insulin, hypertension requiring medication, tobacco smoking within one year of surgery, symptomatic dyspnea (at rest or exertional), history of chronic obstructive pulmonary disease (COPD), systemic infection (SIRS/sepsis/septic shock), wound classification (I: Clean, II-IV: Contaminated or Dirty/Infected), and American Society of Anesthesiology (ASA) class. Laboratory values collected within 30 days of surgery included platelet count (μL−1), white blood cell (WBC) count (μL−1), and hematocrit (HCT; %). Bleeding predisposition was defined as an ongoing hypo-thrombotic condition including hemophilia, vitamin K deficiency, or chronic anticoagulation therapy, but excluding chronic aspirin or NSAID use. Preoperative steroid use was defined as oral or parenteral corticosteroid therapy for a chronic medical condition within 30 days of surgery, excluding a one-time short pulse course bten days in duration. Functional baseline on activities of daily living (ADLs)

was classified as ‘Independent’ or ‘Non-Independent,’ encompassing partially- or fully-dependent conditions. Hospital transfer status was classified as admission from home versus non-home facility (including acute care inpatient, nursing home/chronic care facility, outside emergency department, and others). Patients lacking a documented hospital transfer status were ineligible for study. All continuous variables were converted to ordinal variables as follows: age (65–70 years, 70–80 years, ≥80 years); BMI (underweight, b18.5; normal weight, 18.5–24.9; overweight, 25.0–29.9; obese I, 30.0–34.9; obese II-III, ≥35.0); platelets (thrombocytopenia: b150,000/ μL); WBC count (leukocytosis: ≥12,000/μL); HCT (anemia: b36%); ASA Classification (Class I-II = normal health or mild systemic disease; Class III = severe systemic disease; Class IV-V = life threatening condition or moribund); and operative duration (90–180 min, 180–270 min, ≥270 min). Tumor location was classified using ICD codes as follows (ICD-9/ICD-10): Frontal lobe (191.1/C71.1), Temporal lobe (191.2/ C71.2), Parietal lobe (191.3/C71.3), Occipital lobe (191.4/C71.4), and other/unclassifiable (191.5-9/C71.5-9) (Supplemental Table S1). 2.4. Outcome Measures The primary 30-day outcomes were all-cause mortality, lifethreatening complication, change in living disposition, unplanned readmission, and unplanned reoperation. Life-threatening complications were defined using the Clavien-Dindo Grade IV classification [21,22]: cardiac arrest, myocardial infarction, unplanned re-intubation, postoperative ventilator dependence ≥48 h, septic shock, pulmonary embolism, cerebrovascular accident (CVA), and new renal failure requiring dialysis. Unplanned reoperation and readmission were not limited to the index facility. CPT and ICD codes were used as available to determine the reason for readmission and reoperation. Finally, among patients who were admitted from home for the index surgery, change in living disposition was defined as discharge to a non-home facility or death during the index hospitalization. 2.5. Statistical Analysis Multiple logistic regression was used to assess the relationship between preoperative covariates and postoperative outcomes. Two multiple logistic regressions were performed. First, factors independently associated with a composite outcome of mortality and/or life-threatening complication were identified within the full sample (n = 1016 cases). Second, predictors of change in living disposition were examined among 816 patients who were admitted from home for the surgical admission. Due to the limited number of reoperations and readmissions, these outcomes were reported descriptively. Multiple logistic modeling was performed using the Statistical Analysis Software package (SAS Institute Inc., Cary, NC). Regression parameters were estimated using maximum likelihood estimation (MLE). The adjusted odds ratio and 95% confidence interval (CI) were reported. Significant independent predictors were identified using an α = 0.05 significance threshold. The Hosmer-Lemeshow test was used to evaluate overall model fit (p N 0.05 indicates satisfactory fit). 3. Results 3.1. Demographics and Clinical Characteristics In total, 1016 cases were examined. Case demographics and clinical data are presented in Table 1. The mean age (±Standard Deviation) at time of surgery was 71.9 ± 5.6 years. Common medical co-morbidities included hypertension requiring medication (611, 60.1%), DM on oral agent or insulin (174, 17.1%), and regular use of oral or parenteral corticosteroids near the time of admission (148, 14.6%). The majority of patients were ASA Class III (746, 73.4%). Tumor location was classifiable

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019

R. Rahmani et al. / Journal of Geriatric Oncology xxx (2019) xxx

3

Table 1 Demographics and pre-operative characteristics, 30-day mortality, and major morbidity. Variable

Age 65 ≤ Age b 70 70 ≤ Age b 80 Age ≥ 80 Sex Male Female Race/ethnicity White/Caucasian Black/African American Other race/ethnicity BMI Underweight (b18.5) Normal (18.5–24.9) Overweight (25.0–29.9) Class I Obese (30.0–34.9) Class ≥II Obese (≥35.0) Transfer status Admitted from home Transferred from non-home facility Acute care hospital inpatient Nursing home / chronic care Outside emergency department Baseline functional status Independent Non-independent Partially dependent Totally dependent ASA classification I-II III IV-V Medical co-morbidities DM (Insulin or Oral Agent, REF = No) Anti-hypertensive med use (REF = No) Tobacco smoke (REF = No) History of COPD (REF = No) Dyspnea (Exertional or At Rest, REF = No) Corticosteroid use (REF = No) Bleeding predisposition (REF = No) SIRS/sepsis/septic shock (REF = No) Contaminated/dirty wound (REF = Clean) Laboratory abnormalities Leukocytosis (≥12,000/μL) Anemia (HCT b36%) Thrombocytopenia (b150,000/μL) Tumor location (supratentorial) Frontal lobe Temporal lobe Parietal lobe Occipital lobe Other/unspecified location Operative time 90–180 min 180–270 min ≥270 min

Total (N = 1016)

30-Day mortality and/or life-threatening morbidity

N (%)

N (%)

Adjusted OR (95% CI)a

P-Value

434 (42.7) 463 (45.6) 119 (11.7)

23 (5.3) 41 (8.9) 12 (10.1)

REF 1.64 (0.93–2.88) 2.03 (9.90–4.61)

REF 0.088 0.089

590 (58.1) 426 (41.9)

52 (8.8) 24 (5.6)

REF 0.70 (0.41–1.21)

REF 0.203

797 (78.4) 41 (4.0) 178 (17.5)

61 (7.7) 3 (7.3) 12 (6.7)

REF 1.03 (0.27–3.89) 1.03 (0.52–2.04)

REF 0.967 0.933

14 (1.4) 276 (27.2) 412 (40.6) 211 (20.8) 103 (10.1)

3 (21.4) 15 (5.4) 30 (7.3) 17 (8.1) 11 (10.7)

4.87 (1.05–22.66) REF 1.04 (0.52–2.07) 1.02 (0.47–2.22) 1.44 (0.58–3.59)

0.044 REF 0.910 0.960 0.438

816 (80.3) 200 (19.7) 92 (9.1) 6 (0.6) 102 (10.0)

58 (7.1) 18 (9.0) – – –

REF 1.22 (0.66–2.23) – – –

REF 0.528 – – –

963 (94.8) 53 (5.2) 51 (5.0) 2 (0.2)

66 (6.9) 10 (18.9) – –

REF 2.50 (1.09–5.77) – –

REF 0.031 – –

147 (14.5) 746 (73.4) 123 (12.1)

10 (6.8) 50 (6.7) 16 (13.0)

REF 0.80 (0.37–1.71) 1.46 (0.59–3.66)

REF 0.566 0.415

174 (17.1) 611 (60.1) 87 (8.6) 30 (3.0) 29 (2.9) 148 (14.6) 28 (2.8) 33 (3.2) 34 (3.3)

18 (10.3) 55 (9.0) 12 (13.8) 5 (16.7) 4 (13.8) 19 (12.8) 4 (14.3) 3 (9.1) 1 (2.9)

1.19 (0.63–2.24) 1.50 (0.83–2.73) 2.07 (0.98–4.37) 2.18 (0.72–6.60) 1.77 (0.52–5.98) 1.94 (1.05–3.59) 2.77 (0.89–8.68) 0.58 (0.15–2.33) 0.26 (0.03–2.14)

0.587 0.178 0.056 0.170 0.358 0.036 0.080 0.445 0.212

275 (27.1) 137 (13.5) 101 (9.9)

29 (10.5) 8 (5.8) 12 (11.9)

1.57 (0.91–2.69) 0.60 (0.26–1.37) 1.32 (0.60–2.93)

0.103 0.225 0.488

264 (26.0) 331 (32.6) 162 (15.9) 51 (5.0) 208 (20.5)

26 (9.8) 19 (5.7) 9 (5.6) 0 (0.0) 22 (10.6)

2.19 (1.14–4.24) REF 1.02 (0.43–2.40) – 2.31 (1.16–4.60)

0.019 REF 0.971 – 0.017

542 (53.3) 324 (31.9) 150 (14.8)

41 (7.6) 23 (7.1) 12 (8.0)

REF 1.06 (0.60–1.88) 1.26 (0.61–2.59)

REF 0.842 0.535

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; REF, reference. a Adjusted OR from multiple logistic regression.

in 808 cases (79.5%) and most commonly localized to the temporal (331, 32.6%) or frontal (264, 26.0%) lobes. 3.2. Mortality and Life-Threatening Complication Death within 30-days of surgery occurred in 35 cases (35, 3.4%). The median time to death was sixteen days (Interquartile Range [IQR]: 9–23 days). No deaths occurred within the first three post-operative days. At least one life-threatening complication occurred in 58 cases (5.7%), including: unplanned reintubation (21, 2.1%), ventilator dependence ≥48 h (16, 1.6%), pulmonary embolism (14, 1.4%), and CVA (14,

1.4%). The median time to first life-threatening complication was five days (IQR: 1–18 days), with 18/58 (31.0%) occurring within one day of surgery. The overall rate of the composite outcome of death and/or lifethreatening complication was 7.5%. Rates of the composite outcome increased steadily across age groups (age 65–70 years: 5.3%; age 70–80 years: 8.9%; age ≥ 80 years: 10.1%). Multiple logistic regression identified several independent predictors of composite morbidity and mortality (Table 1). Significant risk factors included (adjusted odds ratio [OR], 95% confidence interval [CI]): non-independent functional status (2.5, 1.1-5.8), frontal tumor location

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019

4

R. Rahmani et al. / Journal of Geriatric Oncology xxx (2019) xxx

(2.2, 1.1-4.2), non-lobar or unclassifiable tumor (2.3, 1.2-4.6), preoperative corticosteroid use (1.9, 1.1-3.6), and underweight BMI (4.9, 1.122.7). 3.3. Change in Living Disposition Most patients were admitted from home (816, 80.3%). Among patients in this group, 277 (33.9%) were discharged to a non-home facility and fifteen (1.8%) expired during the index admission. The most common non-home discharge destinations were rehabilitation (184/816, 22.5%) and skilled care facility (75/816, 9.2%). Four patients admitted from home were discharged to hospice (0.4%). Among patients admitted from home, factors associated with change in living disposition included age ≥ 70 years, female sex, underweight or overweight BMI, non-independent functional status before admission, ASA Class IV-V, systemic infection, DM on oral or insulin therapy, frontal tumor, and

non-lobar or unclassifiable tumor (Table 2). Predictably, patients discharged to a non-home facility were hospitalized for significantly more days during the index admission (median = 7 days, IQR = 7–11 days) compared to patients who returned to home (median = 3 days, IQR = 2–6 days; Wilcoxon rank sum, p = 1.6 × 10−36). The median total length of stay (LOS) across all patients was five days (IQR: 3–8 days). 3.4. Unplanned Reoperation and Readmission Unplanned readmission occurred in 120 cases (11.8%). For half of these cases (61/120, 50.8%), the readmission diagnosis was unspecified. The most common listed reasons for unplanned readmission were altered mental status/encephalopathy (n = 13), seizure/convulsion (n = 12), and venous thromboembolism (n = 11). Unplanned reoperation was less common, occurring in 46 cases (4.5%). In 22 cases (22/46,

Table 2 Rates and predictors of non-routine disposition among patients admitted from home. Variable

Age 65 ≤ Age b 70 70 ≤ Age b 80 Age ≥ 80 Sex Male Female Race/ethnicity White/Caucasian Black/African American Other race/ethnicity BMI Underweight (b18.5) Normal (18.5–24.9) Overweight (25.0–29.9) Class I obese (30.0–34.9) Class ≥II obese (≥35.0) Baseline functional status Independent Non-independent Partially dependent Totally dependent ASA classification I-II III IV-V Medical co-morbidities DM (Insulin or oral agent, REF = No) Anti-hypertensive med use (REF = No) Tobacco smoke (REF = No) History of COPD (REF = No) Dyspnea (Exertional or At rest, REF = No) Corticosteroid use (REF = No) Bleeding predisposition (REF = No) SIRS/sepsis/septic shock (REF = No) Contaminated/dirty wound (REF = Clean) Laboratory abnormalities Leukocytosis (≥12,000/μL) Anemia (HCT b36%) Thrombocytopenia (b150,000/μL) Tumor location (supratentorial) Frontal lobe Temporal lobe Parietal lobe Occipital lobe Other/unspecified location Operative time 90–180 min 180–270 min ≥270 min

Home admit (N = 816)

Non-home disposition

N (%)

N (%)

Adjusted OR (95% CI)a

P-Value

352 (43.1) 370 (45.3) 94 (11.5)

103 (29.3) 142 (38.4) 53 (56.4)

REF 1.42 (1.01–1.98) 3.47 (2.06–5.84)

REF 0.041 b0.0001

473 (58.0) 343 (42.0)

159 (33.6) 139 (40.5)

REF 1.50 (1.08–2.07)

REF 0.015

646 (79.2) 30 (3.7) 140 (17.2)

240 (37.2) 15 (50.0) 43 (30.7)

REF 1.57 (0.69–3.59) 0.72 (0.47–1.11)

REF 0.284 0.139

11 (1.3) 231 (28.3) 338 (41.4) 164 (20.1) 72 (8.8)

7 (63.6) 70 (30.3) 134 (39.6) 60 (36.6) 27 (37.5)

4.24 (1.11–16.28) REF 1.59 (1.07–2.34) 1.19 (0.75–1.91) 0.95 (0.51–1.78)

0.035 REF 0.021 0.464 0.876

777 (95.2) 39 (4.8) 38 (4.7) 1 (0.1)

276 (35.5) 22 (56.4) – –

REF 2.49 (1.20–5.17) – –

REF 0.015 – –

115 (14.1) 613 (75.1) 88 (10.8)

30 (26.1) 228 (37.2) 40 (45.5)

REF 1.47 (0.91–2.38) 2.27 (1.19–4.35)

REF 0.119 0.013

129 (15.8) 479 (58.7) 65 (8.0) 20 (2.5) 23 (2.8) 137 (16.8) 23 (2.8) 16 (2.0) 25 (3.1)

66 (51.2) 194 (40.5) 22 (33.8) 10 (50.0) 10 (43.5) 50 (36.5) 10 (43.5) 11 (68.8) 9 (36.0)

2.00 (1.31–3.06) 1.16 (0.83–1.62) 0.80 (0.44–1.46) 2.07 (0.77–5.53) 1.03 (0.40–2.61) 0.82 (0.53–1.26) 1.24 (0.49–3.16) 3.51 (1.09–11.27) 1.03 (0.42–2.49)

0.002 0.401 0.463 0.149 0.955 0.365 0.652 0.035 0.952

212 (26.0) 103 (12.6) 87 (10.7)

86 (40.6) 48 (46.6) 32 (36.8)

1.17 (0.81–1.68) 1.25 (0.78–2.01) 0.91 (0.54–1.55)

0.410 0.345 0.731

204 (25.0) 268 (32.8) 132 (16.2) 43 (5.3) 169 (20.7)

92 (45.1) 74 (27.6) 50 (37.9) 12 (27.9) 70 (41.4)

2.52 (1.66–3.83) REF 1.49 (0.92–2.41) 1.05 (0.49–2.25) 2.17 (1.40–3.36)

b0.0001 REF 0.101 0.902 b0.001

423 (51.8) 270 (33.1) 123 (15.1)

160 (37.8) 98 (36.3) 40 (32.5)

REF 1.12 (0.79–1.58) 1.07 (0.67–1.72)

REF 0.525 0.773

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; REF, reference. a Adjusted OR from multiple logistic regression.

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019

R. Rahmani et al. / Journal of Geriatric Oncology xxx (2019) xxx

47.8%), the indication for reoperation was unlisted or non-specific. Listed reasons for unplanned reoperation included intracranial hemorrhage (n = 16), surgical site infection (n = 4), and hydrocephalus (n = 4). 4. Discussion We report a retrospective analysis of N1000 older adult patients who underwent craniotomy for resection of primary, malignant, supratentorial, intra-axial brain tumor, documented within the ACSNSQIP national surgical database. Resection improves survival in older adults with malignant glioma, but the decision to undergo surgery must be weighed against the near-term risks imposed by cranial surgery. This study aimed to provide neurosurgeons a clearer picture of the rates and predictors of short-term outcomes after glioma resection in older adult patients to help inform these difficult conversations. 4.1. Motivation for the Current Study In 2013, Seicean and colleagues [14] reported an ACS-NSQIP analysis of older adult patients who underwent craniotomy for malignant tumor resection between the years of 2006–2010. The authors found no differences in adverse 30-day outcome rates between older patients (≥75 years old) and a propensity-matched comparison group (aged 40–74 years). Although this study took an important step towards documenting short-term outcomes from glioma resection in the older adult population, several questions were unanswerable at the time. Notably, ACS-NSQIP began routinely documenting CPT/ICD codes for reoperation and readmissions beginning in 2012, presenting the opportunity to analyze these outcomes descriptively. Similarly, reporting of discharge disposition was not performed during that study's time window, precluding analysis of change in living disposition. Finally, the older cohort examined by Seicean et al. was limited in size (134 patients ≥75 years old compared to 296 patients in this study) and included a combination of primary and secondary tumors. Our study built upon the findings of Seicean et al. in three ways: (1) narrowing our focus to primary brain neoplasms; (2) incorporating a larger sample of older adult patients; and (3) capitalizing on improved ACS-NSQIP documentation of reoperation, readmission, and discharge disposition. 4.2. 30-Day Mortality and Life-Threatening Complication are Rare Outcomes The vast majority of patients (96.6%) were alive 30 days after the index surgery, and relatively few (5.7%) suffered a life-threatening complication in that time span. Surprisingly, rates of mortality and severe morbidity remained fairly low (~10% for the composite outcome) even among the subgroup of patients ≥80 years old. These data suggest that appropriately-selected older adult patients can tolerate the physiological stress imposed by resection of malignant brain tumor with fairly low rates of near-term mortality and severe morbidity, which corroborates previous studies [14,16,23]. Factors independently associated with increased odds of mortality and/or life-threatening complication included two medical comorbidities: functional dependence and low BMI. Impaired functional status, typically quantified by the Karnofsky Performance Status (KPS) scale, is a well-known predictor of poor outcomes from glioma resection in older adults [10,16]. We interpret both impaired performance status and underweight BMI as markers of general medical frailty, or the state of decreased homeostatic and physiological reserve leading to increased vulnerability after surgical stress [24]. In recent years, frailty has been recognized as a common and important risk factor for adverse surgical outcomes for a variety of procedural indications, including tumor resection [25–27]. A study by Youngerman et al. [15] examined an ACS-NSQIP cohort of 9149 patients who underwent surgery for intracranial neoplasms, finding that a composite score of preoperative frailty (including

5

factors such as diabetes mellitus, functional status, congestive heart failure, cerebrovascular disease with neurological deficit, etc.), was associated with higher odds of mortality and other adverse outcomes. Using the same composite score, Cloney et al. [17] measured preoperative frailty in an institutional cohort of older adult patients undergoing GBM resection, finding that frailer patients had an increased rate of complications and decreased overall survival. Our study highlights underweight BMI as an additional risk factor for short-term morbidity and mortality in older patients, which may reflect the contributions of cachexia and/or poor nutritional status to the broader spectrum of frailty. Interestingly, we found that patients taking corticosteroids for a chronic medical condition before surgery had higher odds of morbidity and mortality. The relationship between peri-operative corticosteroid use and surgical outcomes in this population is not well-defined. One ACS-NSQIP-based study of patients who underwent resection for malignant brain tumors (including both metastatic and primary lesions) found that steroids were not independently associated with adverse short-term outcomes [28]. A difficulty in establishing any clear link between corticosteroid use and surgical outcomes using retrospective databases is that results may be confounded by undocumented differences in disease severity and progression between groups. Unfortunately, these important nuances cannot be parsed from the ACS-NSQIP database. This highlights the need to improve and expand existing neurosurgical registries such as The NeuroPoint Registry into the tumor realm. We found that frontal lobe tumor location was an independent risk factor for morbidity and mortality. Frontal lobe tumors may compromise short-term recovery due to disruption of key language, personality, and motor regions, though the lack of database granularity for assessing these neurological outcomes renders our interpretation speculative. Further, ICD postoperative diagnosis codes are unable to capture tumors that infiltrate multiple lobes or cross the midline. The relationship between tumor location and resection outcome has been examined previously, with a study by Chaichana et al. [29] identifying periventricular tumor location as an independent risk factor for poorer survival. Our data indicate that non-classifiable/non-lobar tumor location was also a risk factor for morbidity and mortality, which encompassed ventricular tumors, neoplasms of “overlapping sites,” and otherwise unspecified tumors. 4.3. Change in Living Disposition Occurs in One in Three patients Perhaps the most impactful finding is that one in three patients admitted from home were discharged to a non-home facility after surgery, and discharge to non-home facility was associated with significantly longer index hospitalizations (median of seven days for non-home discharges compared to three days for return-to-home discharges). Bed availability, insurance approval, and care coordination when preparing to discharge patients to a non-home facility all increase the length and cost of the index admission. Beyond the obvious impact on healthcare expense, patients may consider the ~one in three risk of change in living disposition in the context of their individualized goals of care. This possibility should be weighed especially heavily in patients with additional risk factors for change in living disposition, which included a combination of non-modifiable (i.e., age, female sex, tumor location) and potentially modifiable (i.e., BMI, functional status, ASA Class, systemic infection, and DM) characteristics. It is important to note that the nature of malignant glioma undoubtedly will result in a change in living disposition/quality of life during the disease course, and the potential benefits of maximal safe surgical resection should not be overshadowed by a potential discharge to short term rehabilitation. These data also highlight the excellent outcomes possible with a multidisciplinary approach in this high risk patient population undergoing complex intracranial surgery. The rate of non-home discharge observed in our study was slightly higher than reported in a study by Marcus et al. [30], who found that

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019

6

R. Rahmani et al. / Journal of Geriatric Oncology xxx (2019) xxx

24.4% of California patients receiving craniotomy for primary malignant brain tumor resection were discharged to another facility. In a study of malignant glioma resection using the National Inpatient Sample, Missios et al. [31] observed that 25.8% of patients were discharged to a rehabilitation facility. The higher rate observed in our study likely reflects population differences, given that our study focused exclusively on older adult patients. Other findings from the Marcus et al. study generally align with our results, including the overall 30-day mortality rate (2.3%) and median length of stay (six days). 4.4. Unplanned Readmission and Reoperation Readmission within 30-days significantly increases healthcare cost and is used as a measure of care quality [32]. Readmission also exposes patients to the inherent risk of additional time in the hospital environment. In our cohort, we documented an unplanned 30-day readmission rate of ~12%, similar to the readmission rates observed by Marcus et al. [30] (13.2%), Nuño et al. (15.8%), Senders et al. [23] (11.5%), and Donoho et al. [33] (13.2%) for similarly-defined cohorts. The most commonly documented reasons for readmission were altered mental status, seizure, and venous thromboembolism, which generally overlap with the findings of those previous studies. The majority of readmission diagnoses are unlisted in ACS-NSQIP, so it is difficult to rigorously interpret our results. However, these studies generally benchmark 30-day readmission rates after craniotomy for malignant brain tumor resection at 10–15%. Of note, Marcus et al. [30] found that the median hospital charges associated with each 30-day readmission episode were N$20,000, demonstrating the significant impact of readmission on health care costs. Finally, in terms of unplanned 30-day reoperation, the rate observed in this study was fairly low (4.5%) and was most frequently attributed to intracranial hemorrhage (16/46, 34.8%). Other reasons for reoperation included hydrocephalus and surgical site infection. Unfortunately, our ability to interpret this result is limited by the fact that most reoperation procedures were associated with non-specific CPT and ICD codes. The reoperation rate observed here is similar to that reported by Senders et al. [23] (5.1%) in their ACS-NSQIP analysis of resections for primary malignant tumor from years 2005–2015. Seicean et al. [14] also observed a similar reoperation rate (5.7%) in their analysis malignant tumor resections in older patients.

national surgical database, which offers clear advantages in terms of sample size and generalizability. 5. Conclusions We retrospectively examined N1000 cases of older adult patients undergoing resection of primary, malignant, supratentorial, intra-axial tumor. 30-day mortality and life-threatening complication rates were fairly low even among the oldest subgroup in the study (~10%), suggesting that carefully-selected patients can tolerate the near-term risks of cranial surgery. However, 34% of patients admitted from home experienced a change in living disposition postoperatively, with the majority going to rehabilitation (23%) and not a skilled nursing facility (9%). Understanding the rates and risk factors for adverse short-term outcomes after glioma resection should aid neurosurgeons in guiding difficult treatment conversations with patients and families. Supplementary data to this article can be found online at https://doi. org/10.1016/j.jgo.2019.10.019. Funding and Disclosures This work was funded by intramural research support from the Department of Neurosurgery, University of Rochester Medical Center. The authors have no conflicts to disclose. An earlier version of the project was presented as an electronic poster at the 2019 American Association of Neurological Surgeons Annual Meeting, San Diego, CA, April 13–17, 2019. Author Contributions Study Concepts: RR, SBT, GS, KTW, TS, KAW, GEV Study Design: RR, SBT, GS, TS, GEV Data Acquisition: SBT, RR Quality Control of Data and Algorithms: SBT, RR Data Analysis and Interpretation: RR, SBT, TS, GEV Statistical Analysis: SBT, RR, TS, GEV Manuscript Preparation: RR, SBT, GS, KTW, TS, GEV Manuscript Editing: RR, SBT, GS, KTW, TS, KAW, GEV Manuscript Review: RR, SBT, GS, KTW, TS, KAW, GEV

4.5. Limitations

Acknowledgements

There are several limitations of the study. First, all retrospective database studies are subject to contamination of the source data, which can occur due to factors such as erroneous data entry, inaccurate diagnosis coding, and incomplete reporting. The latter issue was particularly evident in our analysis of reoperation and readmission. Second, database research is limited by the use of broad variable definitions that may be suboptimal for the specific question under study. For instance, functional performance was defined to distinguish between patients who were independent on ADLs versus those who exhibited either partial or complete dependence. This definition invites concerns about inter-rater reliability and does not offer the same degree of granularity as more quantitative functional metrics, such as the KPS. Other data elements of interest for this study but unavailable in ACS-NSQIP include genetic tumor markers, extent of tumor resection, socioeconomic status, and type of insurance. The definition of corticosteroid use was also imperfect as it did not allow us to determine whether these medications were prescribed for glioma management or for other chronic conditions (e.g., chronic obstructive pulmonary disease). Additionally, we were unable to identify patients who may have been administered a short course of corticosteroids after resection. Finally, a major opportunity to improve this study is the incorporation of serial radiographic assessments to precisely define tumor location, size, and rate of progression before surgery. However, this would preclude the use of a large

American College of Surgeons National Surgical Quality Improvement Program and the hospitals participating in the ACS-NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors. References [1] Ostrom QT, Gittleman H, Xu J, Kromer C, Wolinsky Y, Kruchko C, et al. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2009-2013. Neuro Oncol 2016;18:v1–v75. [2] Arvold ND, Reardon DA. Treatment options and outcomes for glioblastoma in the elderly patient. Clin Interv Aging 2014;9:357–67. [3] Fisher JL, Schwartzbaum JA, Wrensch M, Wiemels JL. Epidemiology of brain tumors. Neurol Clin 2007;25:867–90 [vii]. [4] Chakrabarti I, Cockburn M, Cozen W, Wang YP, Preston-Martin S. A populationbased description of glioblastoma multiforme in Los Angeles County, 1974-1999. Cancer 2005;104:2798–806. [5] Hoffman S, Propp JM, McCarthy BJ. Temporal trends in incidence of primary brain tumors in the United States, 1985–1999. Neuro Oncol 2006;8:27–37. [6] Dubrow R, Darefsky AS. Demographic variation in incidence of adult glioma by subtype, United States, 1992–2007. BMC Cancer 2011;11:325. [7] Scott JG, Bauchet L, Fraum TJ, Nayak L, Cooper AR, Chao ST, et al. Recursive partitioning analysis of prognostic factors for glioblastoma patients aged 70 years or older. Cancer 2012;118:5595–600. [8] Chaichana KL, Garzon-Muvdi T, Parker S, Weingart JD, Olivi A, Bennett R, et al. Supratentorial glioblastoma multiforme: the role of surgical resection versus biopsy among older patients. Ann Surg Oncol 2011;18:239–45.

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019

R. Rahmani et al. / Journal of Geriatric Oncology xxx (2019) xxx [9] Vuorinen V, Hinkka S, Farkkila M, Jaaskelainen J. Debulking or biopsy of malignant glioma in elderly people-a randomised study. Acta Neurochir 2003;145:5–10. [10] Ewelt C, Goeppert M, Rapp M, Steiger HJ, Stummer W, Sabel M. Glioblastoma multiforme of the elderly: the prognostic effect of resection on survival. J Neurooncol 2011;103:611–8. [11] Tanaka S, Meyer FB, Buckner JC, Uhm JH, Yan ES, Parney IF. Presentation, management, and outcome of newly diagnosed glioblastoma in elderly patients. J Neurosurg 2013;118:786–98. [12] Mohan DS, Suh JH, Phan JL, Kupelian PA, Cohen BH, Barnett GH. Outcome in elderly patients undergoing definitive surgery and radiation therapy for supratentorial glioblastoma multiforme at a tertiary care institution. Int J Radiat Oncol Biol Phys 1998; 42:981–7. [13] Hervey-Jumper SL, Berger MS. Maximizing safe resection of low- and high-grade glioma. J Neurooncol 2016;130:269–82. [14] Seicean A, Seicean S, Schiltz NK, Alan N, Jones PK, Neuhauser D, et al. Short-term outcomes of craniotomy for malignant brain tumors in the elderly. Cancer 2013;119: 1058–64. [15] Youngerman BE, Neugut AI, Yang J, Hershman DL, Wright JD, Bruce JN. The modified frailty index and 30-day adverse events in oncologic neurosurgery. J Neurooncol 2018;136:197–206. [16] D’Amico RS, Cloney MB, Sonabend AM, Zacharia B, Nazarian MN, Iwamoto FM, et al. The safety of surgery in elderly patients with primary and recurrent glioblastoma. World Neurosurg 2015;84:913–9. [17] Cloney M, D’Amico R, Lebovic J, Nazarian M, Zacharia BE, Sisti MB, et al. Frailty in geriatric glioblastoma patients: a predictor of operative morbidity and outcome. World Neurosurg 2016;89:362–7. [18] Sawaya R, Hammoud M, Schoppa D, Hess KR, Wu SZ, Shi WM, et al. Neurosurgical outcomes in a modern series of 400 craniotomies for treatment of parenchymal tumors. Neurosurgery 1998;42:1044–55 [discussion 1055–1046]. [19] Iwamoto FM, Reiner AS, Panageas KS, Elkin EB, Abrey LE. Patterns of care in elderly glioblastoma patients. Ann Neurol 2008;64:628–34. [20] Senders JT, Goldhaber NH, Cote DJ, Muskens IS, Dawood HY, De Vos F, et al. Venous thromboembolism and intracranial hemorrhage after craniotomy for primary malignant brain tumors: a National Surgical quality improvement program analysis. J Neurooncol 2018;136:135–45. [21] Abt NB, Richmon JD, Koch WM, Eisele DW, Agrawal N. Assessment of the predictive value of the modified frailty index for Clavien-Dindo grade IV critical care

[22]

[23]

[24] [25] [26] [27] [28]

[29]

[30]

[31] [32]

[33]

7

complications in major head and neck cancer operations. JAMA Otolaryngol Head Neck Surg 2016;142:658–64. Clavien PA, Barkun J, de Oliveira ML, Vauthey JN, Dindo D, Schulick RD, et al. The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg 2009;250:187–96. Senders JT, Muskens IS, Cote DJ, Goldhaber NH, Dawood HY, Gormley WB, et al. Thirty-day outcomes after craniotomy for primary malignant brain tumors: a national surgical quality improvement program analysis. Neurosurgery 2018;83: 1249–59. Hubbard RE, Story DA. Patient frailty: the elephant in the operating room. Anaesthesia 2014;69(Suppl. 1):26–34. Suskind AM, Finlayson E. A call for frailty screening in the preoperative setting. JAMA Surg 2017;152:240–1. Makary MA, Segev DL, Pronovost PJ, Syin D, Bandeen-Roche K, Patel P, et al. Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 2010;210:901–8. Lin H-S, Watts JN, Peel NM, Hubbard RE. Frailty and post-operative outcomes in older surgical patients: a systematic review. BMC Geriatr 2016;16:157. Alan N, Seicean A, Seicean S, Neuhauser D, Benzel EC, Weil RJ. Preoperative steroid use and the incidence of perioperative complications in patients undergoing craniotomy for definitive resection of a malignant brain tumor. J Clin Neurosci 2015;22: 1413–9. Chaichana K, Parker S, Olivi A, Quinones-Hinojosa A. A proposed classification system that projects outcomes based on preoperative variables for adult patients with glioblastoma multiforme. J Neurosurg 2010;112:997–1004. Marcus LP, McCutcheon BA, Noorbakhsh A, Parina RP, Gonda DD, Chen C, et al. Incidence and predictors of 30-day readmission for patients discharged home after craniotomy for malignant supratentorial tumors in California (1995–2010). J Neurosurg 2014;120:1201–11. Missios S, Kalakoti P, Nanda A, Bekelis K. Craniotomy for glioma resection: a predictive model. World Neurosurg 2015;83:957–64. Dasenbrock HH, Yan SC, Smith TR, Valdes PA, Gormley WB, Claus EB, et al. Readmission after craniotomy for tumor: a national surgical quality improvement program analysis. Neurosurgery 2017;80:551–62. Donoho DA, Wen T, Babadjouni RM, Schwartzman W, Buchanan IA, Cen SY, et al. Predictors of 30- and 90-day readmission following craniotomy for malignant brain tumors: analysis of nationwide data. J Neurooncol 2018;136:87–94.

Please cite this article as: R. Rahmani, S.B. Tomlinson, G. Santangelo, et al., Risk factors associated with early adverse outcomes following craniotomy for malignant glioma in old..., J Geriatr Oncol, https://doi.org/10.1016/j.jgo.2019.10.019