Abstract
Background
We sought to estimate the proportion of patients with cancer treated with immune checkpoint inhibitors (ICI) who die soon after starting ICI in the real world and examine factors associated with early mortality (EM).
Methods
We conducted a retrospective cohort study using linked health administrative data from Ontario, Canada. EM was defined as death from any cause within 60 days of ICI initiation. Patients with melanoma, lung, bladder, head and neck, or kidney cancer treated with ICI between 2012 and 2020 were included.
Results
A total of 7126 patients treated with ICI were evaluated. Fifteen percent (1075 of 7126) died within 60 days of initiating ICI. The highest mortality was observed in patients with bladder and head and neck tumors (approximately 21% each). In multivariable analysis, previous hospital admission or emergency department visit, prior chemotherapy or radiation therapy, stage 4 disease at diagnosis, lower hemoglobin, higher white blood cell count, and higher symptom burden were associated with higher risk of EM. Conversely, patients with lung and kidney cancer (compared with melanoma), lower neutrophil to lymphocytes ratio, and with higher body mass index were less likely to die within 60 days post ICI initiation. In a sensitivity analysis, 30-day and 90-day mortality were 7% (519 of 7126) and 22% (1582 of 7126), respectively, with comparable clinical factors associated with EM identified.
Conclusions
EM is common among patients treated with ICI in the real-world setting and is associated with several patient and tumor characteristics. Development of a validated tool to predict EM may facilitate better patient selection for treatment with ICI in routine practice.
Immune checkpoint inhibitors (ICI) have dramatically altered the landscape of cancer treatment (1-7). Despite the promising outcomes seen with the use of ICI, only a portion of patients derive clinical benefit (8). Therefore, it is crucial to offer these agents to patients most likely to benefit.
Early mortality (EM) following initiation of cancer therapy is well reported in the literature for patients treated with chemotherapy or radiation therapy. However, it is heterogeneously defined and varies substantially, with 30-day and 60-day mortality ranging from 1.7% to 17% and 90-day mortality from 10% to 20% (9-13). This variation is multifactorial and is related to disease, treatment toxicity, and patient-related factors. Disease-related factors include primary cancer site, tumor stage, and treatment characteristics; and patient-related factors include age, sex, and performance status (9-13).
Early cross-over of Kaplan-Meier (KM) survival curves seen in clinical trials of ICI suggests that a subpopulation of patients treated with ICI is at higher risk for EM. Potential confounders include older age and comorbidities that may affect survival (14). Whether EM is the result of disease progression or adverse events associated with ICI is unclear and of interest (15). An accurate estimate of the proportion of patients who die soon after starting ICI and a better understanding of factors associated with EM after ICI use is needed to identify patients at higher risk for EM. This can help identify patients for earlier interventions to avoid serious toxicities, including early death, and select patients who may derive the most benefit from ICI.
In Ontario, Canada’s most populous province with a universal health-care system, ICI were approved for the treatment of patients with select cancers starting in 2012 (Supplementary Table 1, available online). We sought to estimate the proportion of patients who died within 60 days after starting ICI and examine if EM following initiation of ICI is associated with demographic or disease characteristics using a real-world cohort of patients.
Methods
Study design
We conducted a population-based retrospective cohort study using health administrative data held at ICES, an independent, nonprofit research institute that facilitates health services research in Ontario (Supplementary Methods 1, available online). This study adhered to the REporting of studies Conducted using Observational Routinely-collected Data (RECORD) reporting guidelines for studies using routinely collected health data (16).
Data sources
The databases used in this study were deterministically linked using unique encoded identifiers and analyzed at ICES. A detailed description of each database can be found in Supplementary Table 2 (available online).
Population
Eligible patients were adults diagnosed with cancer in Ontario and treated with a least 1 dose of ICI between 2012 and 2020. Cancer diagnosis was identified from the Ontario Cancer Registry database. Receipt of ICI was identified from 2 databases that contain information on systemic therapy delivery (Cancer Activity Level Reporting and New Drug Funding Program). During this timeframe, ICI were approved in Ontario for the treatment of patients with melanoma, lung, bladder, head and neck, and kidney cancers (Supplementary Table 1, available online). Therefore, we only included patients with a diagnosis of one of the cancer sites cited above. Patients with another cancer diagnosis between the diagnosis of the cancer of interest and ICI start or up to 5 years before the cancer diagnosis of interest were excluded to avoid having several systemic therapies given for different cancers during the same time. Codes used to identify eligible patients based on their cancer site and ICI drugs administered are listed in detail in Supplementary Table 3 (available online). ICI included any treatment with atezolizumab, avelumab, durvalumab, ipilimumab, nivolumab, or pembrolizumab or any combination of ICI with other systemic therapies. Drug identification numbers are also presented in Supplementary Table 3 (available online).
Baseline characteristics and other covariates
Patients characteristics and demographics included age at initiation of ICI, sex, income quintile, rurality, regional health authority responsible for administration of public health-care services in Ontario (known as Local Health Integration Network for Ontario; Supplementary Methods 2, available online), cancer center facility level where the ICI were administered with the level of complexity of care delivered and the availability of services differentiating one level from another (level 1 being the most complex; Supplementary Methods 2, available online) (17), Charlson score, Edmonton Symptom Assessment System (ESAS) scores within 60 days before ICI start, laboratory values within 60 days before first ICI (hemoglobin [< or ≥10 g/dL], white blood cells [WBC; < or ≥ 11 000/mm3], platelets [< or ≥350 000/mm3], calcium [< or ≥2.55 mmol/L], creatinine [< or ≥100 µmol/L]), neutrophil to lymphocyte ratio (NLR) (11,18), body mass index (BMI), and hospital admission and emergency department visits within 60 days before ICI start. Disease characteristics consisted of year of diagnosis, tumor site, and whether patients had metastatic disease at diagnosis. Treatment characteristics included prior receipt of chemotherapy with a 1-year lookback or radiation therapy within 60 days from ICI start, intent of ICI treatment, ICI type, ICI dose type (fixed vs weight based), and use of steroids between ICI start and up to 60 days after for patients aged 65 years and older (Ontario Drug Benefit). The details on how each variable was operationalized are shown in Supplementary Table 4 (available online).
Outcomes
The primary outcome was EM defined as death within 60 days after start of ICI. Most ICI are administered every 3 to 4 weeks; as such, we thought it is reasonable to look at EM after more than 1 treatment (2-3 cycles). Because the definition of EM varied in the literature, sensitivity analyses were conducted to assess mortality at 30 and 90 days and by cancer site. In addition, we report on factors associated with EM and 1-year and median overall survival (OS) of all patients and by tumor type, median time on ICI, and median number of treatments administered to patients.
Statistical analysis
Summary statistics were used to describe the study cohort. All continuous variables were reported as means with SD and medians with interquartile ranges (IQR) as appropriate. All categorical variables were reported as frequency counts and proportions. To test for statistical significance between tumor sites, we used 1-way analysis of variance or Kruskal-Wallis tests for continuous variables and χ2 for categorical variables.
Mortality over time overall and by tumor site was estimated using KM curves. OS was defined as the time from ICI start until the date of death or the date of last follow-up if no death occurred. Patients alive at last follow-up were censored.
For the primary outcome, EM was reported for the whole cohort and by tumor site as the proportion of patients who died within 60 days after starting ICI out of all patients treated during the same period. Similarly, for the sensitivity analyses, EM was reported for the whole cohort and by tumor site as the proportion of patients who died at 30 days and 90 days.
A multivariable logistic regression model was used to evaluate factors associated with EM at 30, 60, and 90 days. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were reported. Collinearity and model validity were assessed. Median time on ICI and median number of ICI were estimated as the median time elapsed between ICI start and the date of last ICI in all patients and by tumor site.
Throughout, P values less than or equal to .05 were considered statistically significant. To protect patient privacy, cell sizes less than or equal to 5 are not reported. All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC, USA).
Results
Population characteristics
Among 9162 patients with cancer treated with ICI in Ontario between 2012 and 2020, 7126 met the inclusion criteria and were included in the analysis (Figure 1). Their characteristics are presented in Table 1. The median age of the patients was 67 years (IQR = 60-74 years) and 2971 (42%) were of female sex. More than one-half of the patients had a Charlson Comorbidity index of 0 or 1, and 2565 (36%) had an elevated BMI (≥25). The most common cancer site among the cohort was lung cancer (n = 4142, 58%), followed by melanoma (n = 1705, 24%).
Figure 1.
Cohort creation.
Table 1.
Patients and tumor characteristicsa
| Patient and physician characteristics: full cohort | Total |
|---|---|
| Age at index date, y | |
| Mean (SD) | 66.27 (10.99) |
| Median (IQR) | 67 (60-74) |
| Age <65 | 3178 (44.6%) |
| Age ≥65 | 3948 (55.4%) |
| Age ≥75 | 1644 (23.1%) |
| Age ≥85 | 233 (13.5%) |
| Female, No. (%) | 2971 (41.7%) |
| Income quintile, No. (%) | |
| Quintile 1 | 1382 (19.4%) |
| Quintile 2 | 1511 (21.2%) |
| Quintile 3 | 1417 (19.9%) |
| Quintile 4 | 1419 (19.9%) |
| Quintile 5 | 1397 (19.6%) |
| Rural, yes, No. (%) | 1064 (14.9%) |
| Year of cohort entry, No. (%) | |
| 2012 | 28 (0.4%) |
| 2013 | 78 (1.1%) |
| 2014 | 109 (1.5%) |
| 2015 | 277 (3.9%) |
| 2016 | 734 (10.3%) |
| 2017 | 1257 (17.6%) |
| 2018 | 1973 (27.7%) |
| 2019 | 2670 (37.5%) |
| Geographic region, LHIN, No. (%) | |
| A | 431 (6.0%) |
| B | 632 (8.9%) |
| C | 340 (4.8%) |
| D | 835 (11.7%) |
| E | 325 (4.6%) |
| F | 466 (6.5%) |
| G | 459 (6.4%) |
| H | 749 (10.5%) |
| I | 822 (11.5%) |
| J | 409 (5.7%) |
| K | 735 (10.3%) |
| L | 348 (4.9%) |
| M | 428 (6.0%) |
| N | 147 (2.1%) |
| Tumor site, No. (%) | |
| Head and neck | 233 (3.3%) |
| Melanoma | 1705 (23.9%) |
| Kidney | 757 (10.6%) |
| Bladder | 289 (4.1%) |
| Lung | 4142 (58.1%) |
| ESAS | |
| Missing | 2199 (30.9%) |
| 0 | 746 (10.5%) |
| 1 | 1152 (16.2%) |
| 2 | 913 (12.8%) |
| 3 | 780 (10.9%) |
| 4 | 535 (7.5%) |
| 5 | 427 (6.0%) |
| 6 | 227 (3.2%) |
| 7 | 103 (1.4%) |
| 8 | 27 (0.4%) |
| 9 | 17 (0.2%) |
| 10 | 0 |
| Treating cancer center level, No. (%) | |
| 1 | 3000 (42.1%) |
| 2 | 2233 (31.3%) |
| 3 | 1237 (17.4%) |
| 4 | 610 (8.6%) |
| Other/missing | 46 (0.6%) |
| Radiation therapy, yes, No. (%) | 2040 (28.6%) |
| Chemotherapy, yes, No. (%) | 4777 (67.0%) |
| Hospitalizations before immunotherapy start, yes, No. (%) | 1602 (22.5%) |
| ED visits before immunotherapy start, yes, No. (%) | 1804 (25.3%) |
| Charlson Comorbidity Index | |
| 0 | 2961 (41.6%) |
| 1 | 1090 (15.3%) |
| 2 | 476 (6.7%) |
| 3+ | 377 (5.3%) |
| No hospitalizations | 2222 (31.2%) |
| Immunotherapy drug type, No. (%) | |
| Atezolizumab | 155 (2.2%) |
| Avelumab | 13 (0.2%) |
| Combination | 581 (8.2%) |
| Durvalumab | 422 (5.9%) |
| Ipilimumab | 438 (6.1%) |
| Nivolumab | 2643 (37.1%) |
| Pembrolizumab | 2874 (40.3%) |
| Treatment combination type | |
| ICI alone | 4119 (57.8%) |
| ICI+ICI | 515 (7.2%) |
| ICI + chemo | 2492 (35.0%) |
| Stage 4 cancer at diagnosis, No. (%) | 2676 (37.6%) |
| WBC | |
| >11 000/mm3 | 1071 (15.0%) |
| ≤11 000/mm3 | 4783 (67.1%) |
| No test | 1272 (17.9%) |
| Hb | |
| <10 g/dL | 917 (12.9%) |
| ≥10 g/dL | 5311 (74.5%) |
| No test | 898 (12.6%) |
| Platelets | |
| >350 000/mm3 | 1399 (19.6%) |
| ≤350 000/mm3 | 4768 (66.9%) |
| No test | 959 (13.5%) |
| Calcium | |
| >2.55 mmol/L | 395 (5.5%) |
| ≤2.55 mmol/L | 5094 (71.5%) |
| No test | 1637 (23.0%) |
| Creatinine | |
| >100 µmol/L | 1352 (19.0%) |
| ≤100 µmol/L | 4828 (67.8%) |
| No test | 946 (13.3%) |
| Neutrophil to lymphocyte ratio | |
| Mean (SD) | 7.24 ± 12.00 |
| Median (IQR) | 5 (3-8) |
| No. (%) with data | 5171 (72.6%) |
| Immunotherapy intent | |
| Palliative | 6487 (91.0%) |
| Not palliative | 639 (9.0%) |
| Use of steroids | 1347 (18.9%) |
| Patients aged 65 or older | 1263 (32%) |
| Patient aged <65 y | 84 (2.6%) |
| Use of high-dose steroids (prednisone) ≥60 mg/d | 132 (1.9%) |
| BMI | |
| ≥25 | 2565 (36.0%) |
| <25 | 2035 (28.6%) |
| Missing | 2526 (35.4%) |
| Fixed dose, No. | 731 (10.3%) |
ED visits include any visit to ED whether it led to hospital admission or not, hospital admission includes any admission to the hospital whether it start in the ED or not. BMI = body mass index; ED = emergency department; ESAS = Edmonton Symptom Assessment System; Hb = hemoglobin; ICI = immune checkpoint inhibitor; IQR = interquartile range; LHIN = Local Health Integration Network for Ontario; WBC = white blood cells.
Seventy percent of patients had ESAS summary scores between 0 and 3, and 3000 (42%) patients were treated in a level 1 cancer center. At baseline, 1071 (15%) patients had an elevated WBC count (>11 000/mm3), 917 (13%) had low hemoglobin (<10 g/dL), and 1399 (20%) had elevated platelets (>350 000/mm3). The median NLR was 5 (IQR = 3-8). Within 60 days before ICI start, 1804 (25%) and 1602 (23%) patients were seen in the emergency department and were admitted to the hospital, respectively. Moreover, among patients aged 65 years or more, 1263 (32%) and 2% of patients received steroids and a high dose of steroids within 60 days after ICI start. Baseline characteristics differed between patients depending on their cancer site, including age, sex, ESAS scores, and laboratory values. This is detailed in Supplementary Table 5 (available online).
Early mortality
Overall, 1075 patients (15%) died within 60 days after starting ICI. In a sensitivity analysis, the 30-day mortality was approximately 7% (519), and the 90-mortality was almost 22% (1582). The proportion of patients who died after ICI start at 30, 60, and 90 days is presented in Figure 2. Early mortality varied by tumor site. Patients with bladder and head and neck cancers had the highest early mortality rates, followed by patients with lung cancer, whereas patients with kidney and melanoma had the lowest rates (Figure 3).
Figure 2.
Early mortality at 30, 60, and 90 days in all patients treated with immunotherapy. Survival curve with reference lines at 30, 60, and 90 days.
Figure 3.
Early mortality at 30, 60, and 90 days by tumor site. HNC = head and neck cancer.
Predictors of early mortality
Sixty-day mortality
The unadjusted analysis is presented in Table 2. Characteristics associated with lower early 60-day mortality included older age, lower NLR, and higher BMI, and characteristics associated with a higher risk of death included lung, head and neck, and bladder tumors (compared with melanoma), where the patient received treatment, ED visit and hospital admission, previous radiation and chemotherapy, stage 4 disease at diagnosis, higher Charlson score, ICI with palliative intent, higher ESAS scores, lower hemoglobin, higher WBC, higher platelets, higher creatinine, and concurrent chemotherapy treatment with ICI.
Table 2.
Analyses (unadjusted and adjusted) for 60-day mortalitya
| Predictors of early mortality | Unadjusted analysis |
Adjusted analysis |
||
|---|---|---|---|---|
| OR (95% CI) | P | OR (95% CI) | P | |
| Age | ||||
| Every additional 10 y | 0.92 (0.87 to 0.98) | .005 | 0.96 (0.90 to 1.04) | .31 |
| Sex | ||||
| Reference = “male” | 0.965 (0.84 to 1.09) | .49 | 0.94 (0.81 to 1.10) | .46 |
| Income quintile | ||||
| Reference = “quintile 1” | ||||
| Quintile 2 | 0.82 (0.67 to 1.01) | .06 | 0.84 (0.67 to 1.06) | .13 |
| Quintile 3 | 0.97 (0.79 to 1.20) | .74 | 1.01 (0.80 to 1.27) | .94 |
| Quintile 4 | 0.84 (0.68 to 1.04) | .11 | 0.89 (0.71 to 1.13) | .34 |
| Quintile 5 | 1.06 (0.87 to 1.29) | .59 | 1.19 (0.94 to 1.49) | .14 |
| Rural | ||||
| Ref = Urban | ||||
| Rural | 1.00 (0.84 to 1.20) | .96 | 1.00 (0.80 to 1.26) | 1.00 |
| Year of cohort entry | ||||
| Index year, per year | 0.96 (0.91 to 1.00) | .06 | 0.99 (0.92 to 1.06) | .69 |
| LHIN | ||||
| Ref = G | ||||
| A | 0.99 (0.68 to 1.47) | .98 | 1.04 (0.64 to 1.68) | .89 |
| B | 1.00 (0.70 to 1.43) | 1.00 | 1.02 (0.67 to 1.54) | .94 |
| C | 1.21 (0.81 to 1.79) | .36 | 1.18 (0.73 to 1.91) | .50 |
| D | 1.24 (0.89 to 1.71) | .20 | 1.32 (0.90 to 1.94) | .16 |
| E | 1.63 (1.11 to 2.39) | .01 | 1.34 (0.86 to 2.09) | .20 |
| F | 0.84 (0.57 to 1.24) | .38 | 0.95 (0.61 to 1.49) | .83 |
| H | 1.33 (0.96 to 1.85) | .09 | 1.29 (0.89 to 1.88) | .18 |
| I | 1.05 (0.75 to 1.47) | .77 | 0.94 (0.62 to 1.45) | .79 |
| J | 1.37 (0.95 to 1.99) | .10 | 1.71 (1.10 to 2.67) | .02 |
| K | 1.35 (0.97 to 1.88) | .08 | 1.55 (1.04 to 2.31) | .03 |
| L | 1.02 (0.68 to 1.53) | .93 | 0.93 (0.56 to 1.54) | .77 |
| M | 1.17 (0.80 to 1.70) | .42 | 1.04 (0.66 to 1.63) | .88 |
| N | 0.97 (0.56 to 1.68) | .91 | 0.81 (0.42 to 1.56) | .53 |
| Tumor site | ||||
| Per site (reference melanoma) | ||||
| Lung | 1.49 (1.26 to 1.76) | <.001 | 0.63 (0.49 to 0.82) | <.001 |
| Kidney | 0.88 (0.67 to 1.16) | .38 | 0.47 (0.33 to 0.66) | <.001 |
| HNC | 1.94 (1.37 to 2.76) | <.001 | 0.90 (0.58 to 1.39) | .63 |
| Bladder | 1.96 (1.42 to 2.70) | <.001 | 0.78 (0.49 to 1.23) | .28 |
| Cancer center level | ||||
| Ref = (level 1) | ||||
| 2 | 1.05 (0.90 to 1.22) | .56 | 1.38 (1.10 to 1.74) | .006 |
| 3 | 1.38 (1.15 to 1.64) | <.001 | 1.28 (1.01 to 1.62) | .04 |
| 4 | 1.12 (0.88 to 1.42) | .38 | 1.21 (0.89 to 1.64) | .22 |
| Missing | 1.10 (0.49 to 2.47) | .82 | 1.36 (0.55 to 3.37) | .50 |
| Charlson score | ||||
| Ref = 0 | ||||
| No hosp | 0.81 (0.59 to 1.11) | .19 | 0.87 (0.62 to 1.23) | .44 |
| 1 | 0.91 (0.55 to 1.51) | .71 | 0.78 (0.45 to 1.36) | .38 |
| 2 | 0.95 (0.66 to 1.37) | .77 | 0.78 (0.52 to 1.16) | .22 |
| 3+ | 2.03 (1.51 to 2.73) | <.001 | 1.16 (0.83 to 1.61) | .39 |
| ED visit before immunotherapy start | ||||
| Yes vs no | 2.37 (2.07 to 2.71) | <.001 | 1.53 (1.31 to 1.79) | <.001 |
| Hospitalization before immunotherapy start | ||||
| Yes vs no | 3.46 (3.02 to 3.97) | <.001 | 1.95 (1.63 to 2.33) | <.001 |
| Radiation therapy | ||||
| Yes vs no | 1.78 (1.55 to 2.04) | <.001 | 1.36 (1.16 to 1.59) | <.001 |
| Chemotherapy | ||||
| Yes vs no | 1.22 (1.06 to 1.40) | .007 | 1.36 (1.12 to 1.64) | .002 |
| De novo stage 4 cancer | ||||
| Yes vs no | 1.71 (1.50 to 1.95) | <.001 | 1.28 (1.10 to 1.50) | .002 |
| Intent of immunotherapy | ||||
| Palliative vs no | 41 961 (13.47 to 130.69) | <.001 | 21.95 (6.88 to 70.09) | <.001 |
| ESAS | ||||
| ref = 0 | ||||
| 1 | 1.35 (0.91 to 2.02) | .14 | 1.03 (0.67 to 1.56) | .91 |
| 2 | 3.37 (2.32 to 4.90) | <.001 | 2.21 (1.49 to 3.28) | <.001 |
| 3 | 3.79 (2.61 to 5.53) | <.001 | 2.25 (1.51 to 3.36) | <.001 |
| 4 | 5.74 (3.92 to 8.42) | <.001 | 3.54 (2.35 to 5.33) | <.001 |
| 5 | 6.15 (4.15 to 9.12) | <.001 | 3.20 (2.10 to 4.90) | <.001 |
| 6 | 6.39 (4.11 to 9.95) | <.001 | 3.26 (2.02 to 5.25) | <.001 |
| 7 | 8.78 (5.19 to 14.88) | <.001 | 4.36 (2.44 to 7.76) | <.001 |
| 8 | 12 809 (5.56 to 29.50) | <.001 | 7.37 (2.95 to 18.43) | <.001 |
| 9 | 13.04 (4.71 to 36.15) | <.001 | 7.25 (2.22 to 23.70) | .001 |
| Missing | 3.48 (2.46 to 4.92) | <.001 | 2.48 (1.71 to 3.59) | <.001 |
| Hemoglobin | ||||
| <10 vs ≥10 g/dL | 2.56 (2.17 to 3.02) | <.001 | 1.62 (1.34 to 1.97) | <.001 |
| Missing | 1.23 (1.01 to 1.50) | .04 | 1.04 (0.51 to 2.12) | .91 |
| WBC | ||||
| >11 000 vs ≤11 000/mm3 | 4.20 (3.59 to 4.91) | <.001 | 2.38 (1.98 to 2.86) | <.001 |
| Missing | 1.51 (1.27 to 1.81) | <.001 | 1.65 (1.07 to 2.54) | .02 |
| Platelets | ||||
| >350 000 vs ≤350 000/mm3 | 1.69 (1.45 to 1.97) | <.001 | 0.89 (0.74 to 1.07) | .21 |
| Missing | 1.16 (0.96 to 1.41) | .14 | 1.31 (0.63 to 2.74) | .48 |
| Calcium | ||||
| >2.55 vs ≤2.55 mmol/L | 1.11 (0.85 to 1.46) | .45 | 1.17 (0.86 to 1.60) | .32 |
| Missing | 0.76 (0.65 to 0.90) | .001 | 0.73 (0.55 to 0.97) | .03 |
| Creatinine | ||||
| >100 vs ≤100 µmol/L | 0.82 (0.69 to 0.98) | .03 | 0.91 (0.74 to 1.13) | .39 |
| Missing | 0.92 (0.75 to 1.12) | .39 | 0.91 (0.55 to 1.49) | .70 |
| Neutrophil to lymphocyte ratio | ||||
| < vs > Median (4.64) | 0.21 (0.17 to 0.25) | <.001 | 0.36 (0.29 to 0.44) | <.001 |
| Missing | 0.45 (0.39 to 0.53) | <.001 | 0.56 (0.43 to 0.74) | <.001 |
| BMI | ||||
| Ref ≤25 | 0.64 (0.55 to 0.75) | <.001 | ||
| >25 | 0.75 (0.63 to 0.90) | .002 | ||
| Missing | 0.86 (0.70 to 1.05) | .15 | ||
| Concurrent chemotherapy | ||||
| Ref = no combination (single ICI) | 1.37 (1.20 to 1.56) | <.001 | 0.68 (0.47 to 0.99) | .04 |
| ICI + ICI | ||||
| ICI + chemo | 1.09 (0.90 to 1.31) | .37 | ||
| Dose | ||||
| Fixed dose vs not | 1.05 (0.85 to 1.29) | .68 | 1.06 (0.82 to 1.38) | .66 |
BMI = body mass index; CI = confidence interval; ED = emergency department; ESAS = Edmonton Symptom Assessment System; HNC = head and neck cancer; ICI = immune checkpoint inhibitor; IQR = interquartile range; LHIN = Local Health Integration Network for Ontario; WBC = white blood cells.
In the adjusted model presented in Table 2, patients treated in a level 2 or 3 cancer center compared with a level 1 center had a higher risk of mortality. Patients who visited an ED (OR = 1.53, 95% CI = 1.31 to 1.79) or were admitted to a hospital (OR = 1.95, 95% CI = 1.63 to 2.33) were more likely to die at 60 days post ICI initiation. Similarly, patients who were treated with radiation 60 days before ICI start (OR = 1.36, 95% CI = 1.16 to 1.59) and patients who received chemotherapy within 1 year before ICI start (OR = 1.36, 95% CI = 1.12 to 1.64) or with stage 4 cancer at diagnosis (OR = 1.28, 95% CI = 1.10 to 1.50) were at higher risk from dying within 60 days from ICI start. Likewise, patients with higher ESAS scores, lower hemoglobin (OR = 1.62, 95% CI = 1.34 to 1.97), and higher WBC (OR = 2.38, 95% CI = 1.98 to 2.86) were at higher risk from dying early after starting ICI.
Patients with lung and kidney tumors (OR = 0.63, 95% CI = 0.49 to 0.82 and OR = 0.47, 95% CI = 0.33 to 0.66, respectively) were less likely to die at 60 days compared with patients with melanoma. Patients with lower NLR (OR = 0.36, 95% CI = 0.29 to 0.44) and higher BMI (OR = 0.75, 95% CI = 0.63 to 0.90) and those who received ICI + ICI (compared with ICI alone) (OR = 0.68, 95% CI = 0.47 to 0.99) were less likely to die after ICI start as well. The c-statistics for the model was 0.89 (Supplementary Figure 1, available online).
Sensitivity analysis: 30-day mortality
The unadjusted and adjusted analyses are presented in Table 3. The same factors associated with 60-day early mortality were identified except for the cancer center level, lung cancer site, prior chemotherapy, BMI, and stage 4 disease at diagnosis.
Table 3.
Analyses (unadjusted and adjusted) of early mortality at 30 daysa
| Predictors of early mortality | Unadjusted analysis |
Adjusted analysis |
||
|---|---|---|---|---|
| OR (95% CI) | P | OR (95% CI) | P | |
| Age | ||||
| Every additional 10 y | 0.92 (0.85 to 0.10) | .04 | 0.94 (0.86 to 1.03) | .20 |
| Sex | ||||
| Reference = “male” | 0.95 (0.79 to 1.14) | .56 | 0.95 (0.78 to 1.16) | .61 |
| Income quintile | ||||
| Reference = “quintile 1” | ||||
| Quintile 2 | 0.79 (0.59 to 1.05) | .11 | 0.80 (0.59 to 1.10) | .15 |
| Quintile 3 | 1.07 (0.82 to 1.41) | .61 | 1.11 (0.83 to 1.50) | .48 |
| Quintile 4 | 0.87 (0.66 to 1.16) | .36 | 0.91 (0.67 to 1.25) | .57 |
| Quintile 5 | 1.01 (0.76 to 1.33) | .95 | 1.10 (0.81 to 1.49) | .54 |
| Rural | ||||
| Ref = urban | ||||
| Rural | 0.94 (0.73 to 1.22) | .66 | 0.94 (0.69 to 1.28) | .69 |
| Year of cohort entry | ||||
| Index year, per year | 1.02 (0.95 to 1.09) | .60 | 1.05 (0.95 to 1.15) | .34 |
| LHIN | ||||
| Ref = G | ||||
| A | 0.97 (0.58 to 1.61) | .89 | 0.95 (0.50 to 1.78) | .87 |
| B | 1.01 (0.64 to 1.61) | .96 | 1.00 (0.59 to 1.71) | .99 |
| C | 1.07 (0.63 to 1.82) | .81 | 0.99 (0.53 to 1.87) | .99 |
| D | 1.11 (0.72 to 1.71) | .64 | 1.09 (0.66 to 1.79) | .74 |
| E | 1.22 (0.72 to 2.06) | .46 | 0.92 (0.51 to 1.66) | .78 |
| F | 0.83 (0.49 to 1.39) | .47 | 1.05 (0.59 to 1.87) | .87 |
| H | 1.17 (0.75 to 1.81) | .50 | 1.11 (0.68 to 1.81) | .67 |
| I | 0.77 (0.48 to 1.22) | .26 | 0.76 (0.43 to 1.35) | .35 |
| J | 1.13 (0.69 to 1.87) | .63 | 1.28 (0.71 to 2.23) | .41 |
| K | 1.17 (0.75 to 1.82) | .49 | 1.24 (0.74 to 2.07) | .42 |
| L | 0.79 (0.44 to 1.40) | .41 | 0.75 (0.38 to 1.49) | .41 |
| M | 0.94 (0.56 to 1.57) | .81 | 0.79 (0.43 to 1.45) | .45 |
| N | 1.04 (0.51 to 2.12) | .91 | 0.84 (0.37 to 1.91) | .67 |
| Tumor site | ||||
| Per site (reference melanoma) | ||||
| Lung | 1.67 (1.31 to 2.13) | <.001 | 0.81 (0.57 to 1.15) | .24 |
| Kidney | 0.92 (0.62 to 1.37) | .67 | 0.53 (0.33 to 0.86) | .01 |
| HNC | 2.21 (1.38 to 3.52) | <.001 | 1.07 (0.61 to 1.88) | .82 |
| Bladder | 1.74 (1.10 to 2.76) | .02 | 0.65 (0.35 to 1.22) | .18 |
| Cancer center level | ||||
| Ref = (level 1) | ||||
| 2 | 0.94 (0.76 to 1.17) | .59 | 1.20 (0.88 to 1.64) | .25 |
| 3 | 1.36 (1.07 to 1.72) | .01 | 1.18 (0.87 to 1.60) | .29 |
| 4 | 1.09 (0.78 to 1.52) | .61 | 1.11 (0.74 to 1.67) | .60 |
| Missing | 0.93 (0.29 to 3.03) | .91 | 1.09 (0.30 to 3.88) | .90 |
| Charlson score | ||||
| Ref = 0 | ||||
| No hosp | 0.84 (0.52 to 1.35) | .46 | 0.92 (0.56 to 1.51) | .74 |
| 1 | 0.45 (0.17 to 1.20) | .11 | 0.35 (0.13 to 0.96) | .04 |
| 2 | 1.00 (0.58 to 1.71) | .99 | 0.80 (0.45 to 1.42) | .44 |
| 3+ | 2.43 (1.56 to 3.78) | <.001 | 1.34 (0.83 to 2.15) | .23 |
| ED visit before immunotherapy starts | ||||
| Yes vs no | 2.46 (2.05 to 2.95) | <.001 | 1.56 (1.28 to 1.90) | <.001 |
| Hospitalization before immunotherapy start | ||||
| Yes vs no | 4.01 (3.34 to 4.81) | <.001 | 1.99 (1.58 to 2.52) | <.001 |
| Radiation therapy | ||||
| Yes vs no | 1.82 (1.52 to 2.19) | <.001 | 1.26 (1.02 to 1.56) | .03 |
| Chemotherapy | ||||
| Yes vs no | 0.98 (0.81 to 1.19) | .85 | 1.08 (0.84 to 1.38) | .56 |
| De novo stage 4 cancer | ||||
| Yes vs no | 1.55 (1.30 to 1.85) | <.001 | 1.07 (0.87 to 1.31) | .53 |
| Intent of immunotherapy | ||||
| Palliative vs no | 27.58 (6.86 to 110.84) | <.001 | 15.92 (3.83 to 66.21) | <.001 |
| ESAS | ||||
| Ref = 0 | ||||
| 1 | 1.59 (0.85 to 2.98) | .15 | 1.21 (0.63 to 2.31) | .57 |
| 2 | 3.22 (1.77 to 5.85) | <.001 | 1.96 (1.06 to 3.64) | .03 |
| 3 | 4.28 (2.37 to 7.73) | <.001 | 2.38 (1.29 to 4.40) | .006 |
| 4 | 6.98 (3.87 to 12.60) | <.001 | 3.92 (2.12 to 7.25) | <.001 |
| 5 | 7.57 (4.15 to 13.80) | <.001 | 3.68 (1.96 to 6.89) | <.001 |
| 6 | 8.58 (4.49 to 16.40) | <.001 | 4.10 (2.08 to 8.09) | <.001 |
| 7 | 10.34 (4.92 to 21.70) | <.001 | 4.38 (1.99 to 9.63) | <.001 |
| 8 | 11.88 (3.93 to 35.90) | <.001 | 5.94 (1.81 to 19.55) | .003 |
| 9 | 11.20 (2.89 to 43.41) | <.001 | 5.20 (1.20 to 22.64) | .03 |
| Missing | 4.80 (2.77 to 8.32) | <.001 | 3.09 (1.75 to 5.48) | <.001 |
| Hemoglobin | ||||
| <10 vs ≥10 g/dL | 2.32 (1.86 to 2.90) | <.001 | 1.48 (1.15 to 1.91) | .002 |
| Missing | 1.24 (0.95 to 1.63) | .12 | 1.08 (0.41 to 2.85) | .88 |
| WBC | ||||
| >11 000 vs ≤11000/mm3 | 4.33 (3.53 to 5.30) | <.001 | 2.30 (1.82 to 2.90) | <.001 |
| Missing | 1.48 (1.15 to 1.90) | .002 | 1.39 (0.75 to 2.57) | .29 |
| Platelets | ||||
| >350 000 vs ≤350 000/mm3 | 1.49 (1.20 to 1.84) | <.001 | 0.79 (0.62 to 1.01) | .06 |
| Missing | 1.13 (0.87 to 1.48) | .37 | 2.05 (0.72 to 5.81) | .18 |
| Calcium | ||||
| >2.55 vs ≤2.55 mmol/L | 1.08 (0.74 to 1.57) | .71 | 1.15 (0.76 to 1.75) | .50 |
| Missing | 0.80 (0.64 to 1.01) | .06 | 0.86 (0.59 to 1.26) | .44 |
| Creatinine | ||||
| >100 vs ≤100 umol/L | 0.81 (0.64 to 1.04) | .10 | 0.96 (0.72 to 1.27) | .75 |
| Missing | 0.91 (0.69 to 1.19) | .49 | 0.72 (0.36 to 1.45) | .36 |
| Neutrophil to lymphocyte ratio | ||||
| < vs > Median (4.64) | 0.18 (0.14 to 0.24) | <.001 | 0.34 (0.26 to 0.46) | <.001 |
| Missing | 0.44 (0.35 to 0.55) | <.001 | 0.50 (0.35 to 0.73) | <.001 |
| BMI | ||||
| Ref: <25 | 0.71 (0.57 to 0.88) | .002 | ||
| >25 | 0.87 (0.69 to 1.11) | .27 | ||
| Missing | 1.027 (0.79 to 1.34) | .85 | ||
| Concurrent chemo | ||||
| Ref: no combination (single ICI) | 1.20 (1.00 to 1.44) | .05 | ||
| ICI + ICI | 0.56 (0.34 to 0.95) | .03 | ||
| ICI + chemo | 1.02 (0.79 to 1.30) | .90 | ||
| Dose | ||||
| Fixed dose vs not | 1.06 (0.80 to 1.42) | .68 | 1.07 (0.75 to 1.51) | .72 |
BMI = body mass index; CI = confidence interval; ED = emergency department; ESAS = Edmonton Symptom Assessment System; HNC = head and neck cancer; ICI = immune checkpoint inhibitor; IQR = interquartile range; LHIN = Local Health Integration Network for Ontario; WBC = white blood cells.
Sensitivity analysis: 90-day mortality
The unadjusted and adjusted models for the 90-day mortality are presented in Table 4. The same variables were identified as associated with EM in addition to sex (female sex associated with lower risk of mortality) and treatment combination with chemotherapy and ICI (chemotherapy + ICI associated with high risk of mortality compared with single-agent ICI).
Table 4.
Analyses (unadjusted and adjusted) of early mortality at 90 daysa
| Predictors of early mortality | Unadjusted analysis |
Adjusted analysis |
||
|---|---|---|---|---|
| OR (95% CI) | P | OR (95% CI) | P | |
| Age | ||||
| Every additional 10 y | 0.95 (0.90 to 1.00) | .05 | 1.00 (0.94 to 1.06) | .98 |
| Sex | ||||
| Reference = “Male” | 0.88 (0.78 to 0.98) | .03 | 0.85 (0.74 to 0.97) | .01 |
| Income quintile | ||||
| Reference = “Quintile 1” | ||||
| Quintile 2 | 0.91 (0.76 to 1.09) | .29 | 0.94 (0.77 to 1.14) | .54 |
| Quintile 3 | 0.94 (0.78 to 1.12) | .47 | 0.99 (0.81 to 1.20) | .88 |
| Quintile 4 | 0.93 (0.78 to 1.11) | .41 | 1.01 (0.82 to 1.23) | .93 |
| Quintile 5 | 1.05 (0.88 to 1.25) | .61 | 1.18 (0.97 to 1.44) | .10 |
| Rural | ||||
| Ref = Urban | ||||
| Rural | 0.97 (0.83 to 1.14) | .74 | 0.94 (0.77 to 1.15) | .55 |
| Year of cohort entry | ||||
| Index year, per year | 0.95 (0.92 to 0.99) | .02 | 0.97 (0.91 to 1.02) | .24 |
| LHIN | ||||
| Ref = G | ||||
| A | 0.93 (0.67 to 1.29) | .66 | 1.01 (0.67 to 1.53) | .95 |
| B | 0.96 (0.71 to 1.29) | .79 | 1.03 (0.73 to 1.47) | .86 |
| C | 1.22 (0.87 to 1.71) | .24 | 1.30 (0.86 to 1.96) | .21 |
| D | 1.08 (0.82 to 1.43) | .57 | 1.15 (0.82 to 1.60) | .42 |
| E | 1.32 (0.94 to 1.84) | .11 | 1.08 (0.73 to 1.59) | .71 |
| F | 0.84 (0.61 to 1.17) | .31 | 0.93 (0.64 to 1.35) | .70 |
| H | 1.21 (0.92 to 1.61) | .18 | 1.14 (0.83 to 1.57) | .43 |
| I | 1.02 (0.77 to 1.34) | .92 | 0.89 (0.62 to 1.28) | .52 |
| J | 1.22 (0.89 to 1.68) | .21 | 1.55 (1.06 to 2.27) | .03 |
| K | 1.21 (0.91 to 1.60) | .19 | 1.37 (0.97 to 1.93) | .08 |
| L | 0.94 (0.66 to 1.32) | .71 | 0.90 (0.58 to 1.39) | .64 |
| M | 1.18 (0.86 to 1.62) | .30 | 1.13 (0.77 to 1.66) | .55 |
| N | 0.97 (0.61 to 1.54) | .90 | 0.84 (0.49 to 1.47) | .55 |
| Tumor site | ||||
| Per site (reference melanoma) | ||||
| Lung | 1.53 (1.32 to 1.77) | <.001 | 0.69 (0.55 to 0.87) | .001 |
| Kidney | 1.07 (0.85 to 1.33) | .58 | 0.61 (0.46 to 0.81) | <.001 |
| HNC | 1.95 (1.43 to 2.65) | <.001 | 0.99 (0.68 to 1.46) | .98 |
| Bladder | 2.11 (1.59 to 2.79) | <.001 | 0.95 (0.64 to 1.41) | .79 |
| Cancer center level | ||||
| Ref = (level 1) | ||||
| 2 | 1.04 (0.91 to 1.19) | .56 | 1.27 (1.04 to 1.55) | .02 |
| 3 | 1.37 (1.17 to 1.59) | <.001 | 1.24 (1.01 to 1.52) | .04 |
| 4 | 1.18 (0.96 to 1.46) | .11 | 1.19 (0.91 to 1.54) | .20 |
| Missing | 1.86 (0.10 to 3.46) | .05 | 2.11 (1.01 to 4.39) | .05 |
| Charlson score | ||||
| Ref = 0 | ||||
| No hosp | 0.81 (0.62 to 1.05) | .11 | 0.87 (0.65 to 1.16) | .35 |
| 1 | 1.02 (0.68 to 1.53) | .94 | 0.92 (0.59 to 1.44) | .72 |
| 2 | 0.89 (0.66 to 1.20) | .44 | 0.74 (0.53 to 1.03) | .08 |
| 3+ | 1.87 (1.46 to 2.40) | <.001 | 1.12 (0.84 to 1.48) | .44 |
| ED visit before immunotherapy starts | ||||
| Yes vs no | 2.26 (2.01 to 2.55) | <.001 | 1.51 (1.31 to 1.73) | <.001 |
| Hospitalization before immunotherapy start | ||||
| Yes vs no | 3.01 (2.66 to 3.40) | <.001 | 1.80 (1.53 to 2.11) | <.001 |
| Radiation therapy | ||||
| Yes vs no | 1.80 (1.60 to 2.02) | <.001 | 1.56 (1.35 to 1.79) | <.001 |
| Chemotherapy | ||||
| Yes vs no | 1.42 (1.25 to 1.61) | <.001 | 1.57 (1.33 to 1.85) | <.001 |
| De novo stage 4 cancer | ||||
| Yes vs no | 1.61 (1.44 to 1.80) | <.001 | 1.19 (1.04 to 1.37) | .01 |
| Intent of immunotherapy | ||||
| Palliative vs no | 20.11 (10.74 to 37.65) | <.001 | 10.34 (5.34 to 20.02) | <.001 |
| ESAS | ||||
| Ref = 0 | ||||
| 1 | 1.45 (1.06 to 1.99) | .02 | 1.12 (0.80 to 1.57) | .51 |
| 2 | 3.01 (2.22 to 4.09) | <.001 | 2.02 (1.46 to 2.79) | <.001 |
| 3 | 3.98 (2.94 to 5.40) | <.001 | 2.46 (1.77 to 3.41) | <.001 |
| 4 | 5.45 (3.97 to 7.48) | <.001 | 3.46 (2.46 to 4.87) | <.001 |
| 5 | 6.16 (4.44 to 8.53) | <.001 | 3.37 (2.36 to 4.80) | <.001 |
| 6 | 5.66 (3.88 to 8.26) | <.001 | 2.99 (1.98 to 4.52) | <.001 |
| 7 | 7.00 (4.36 to 11.24) | <.001 | 3.82 (2.26 to 6.44) | <.001 |
| 8 | 11.87 (5.34 to 26.38) | <.001 | 7.35 (3.08 to 17.56) | <.001 |
| 9 | 15.75 (5.79 to 42.81) | <.001 | 10.35 (3.26 to 32.90) | <.001 |
| Missing | 3.35 (2.53 to 4.42) | <.001 | 2.52 (1.87 to 3.41) | <.001 |
| Hemoglobin | ||||
| <10 vs ≥ 10 g/dL | 2.51 (2.16 to 2.91) | <.001 | 1.54 (1.29 to 1.83) | <.001 |
| Missing | 1.12 (0.94 to 1.34) | .19 | 0.99 (0.54 to 1.82) | .97 |
| WBC | ||||
| >11 000 vs ≤ 11 000/mm3 | 3.59 (3.12 to 4.14) | <.001 | 2.12 (1.79 to 2.50) | <.001 |
| Missing | 1.34 (1.15 to 1.56) | <.001 | 1.72 (1.19 to 2.50) | .004 |
| Platelets | ||||
| >350 000 vs ≤ 350 000/mm3 | 1.72 (1.51 to 1.97) | <.001 | 0.99 (0.84 to 1.16) | .88 |
| Missing | 1.08 (0.91 to 1.28) | .39 | 1.13 (0.60 to 2.14) | .71 |
| Calcium | ||||
| >2.55 vs ≤ 2.55 mmol/L | 1.15 (0.91 to 1.45) | .25 | 1.18 (0.90 to 1.55) | .23 |
| Missing | 0.77 (0.67 to 0.89) | <.001 | 0.85 (0.68 to 1.08) | .19 |
| Creatinine | ||||
| >100 vs ≤ 100 umol/L | 0.89 (0.76 to 1.03) | .10 | 0.91 (0.76 to 1.09) | .32 |
| Missing | 0.86 (0.72 to 1.02) | .07 | 0.77 (0.50 to 1.18) | .23 |
| Neutrophil to lymphocyte ratio | ||||
| < vs > median (4.64) | 0.25 (0.21 to 0.28) | <.001 | 0.40 (0.34 to 0.47) | <.001 |
| Missing | 0.45 (0.39 to 0.51) | <.001 | 0.57 (0.45 to 0.71) | <.001 |
| BMI | ||||
| Ref = <25 | 0.69 (0.60 to 0.79) | <.001 | ||
| >25 | 0.82 (0.70 to 0.96) | .011 | ||
| Missing | 0.75 (0.63 to 0.90) | .002 | ||
| Concurrent chemo | ||||
| Ref = no combination (single ICI) | 1.54 (1.37 to 1.72) | <.001 | ||
| ICI + ICI | 0.79 (0.58 to 1.08) | .14 | ||
| ICI + chemo | 1.25 (1.06 to 1.47) | .008 | ||
| Dose | ||||
| Fixed dose vs not | 1.03 (0.86 to 1.24) | .73 | 1.05 (0.83 to 1.32) | .69 |
BMI = body mass index; CI = confidence interval; ED = emergency department; ESAS = Edmonton Symptom Assessment System; HNC = head and neck cancer; ICI = immune checkpoint inhibitor; IQR = interquartile range; LHIN = Local Health Integration Network for Ontario; WBC = white blood cells.
Survival outcomes
The median OS for the entire cohort was 11.6 months (Supplementary Figure 2, available online), and median OS by tumor site varied from 6.5 months (for bladder cancer) to 17.3 months (for melanoma) and is presented in Figure 4. As for the 1-year survival, 49% of patients were alive at 1 year. One-year survival by tumor site is presented in Supplementary Table 6 (available online).
Figure 4.
Median overall survival (OS) by tumor site.
Treatment characteristics
The median number of treatments was 4 or 5 depending on tumor site, and the mean, median time and range on ICI varied from 56 to 77 days (Supplementary Table 7, available online).
Discussion
In our population-based study, we found that EM is common among patients with solid tumors treated with ICI: EM was 15% at 60 days and 7% and 22% at 30 and 90 days, respectively. This is in line with previous real-world data on EM post initiation of radiation or chemotherapy. In a large study of patients receiving palliative radiotherapy, 17% died 30 days after starting palliative radiotherapy (9); in another study 8% of patients died 30 days after starting chemotherapy, with most deaths occurring in patients with noncurable disease (18). Our results are also comparable with recent small real-world data reports on EM in patients treated with ICI. A Canadian study showed that the EM at 60 days was 15% in patients with lung cancer (N = 339) (19), and an Australian study (N = 601) showed that EM was 12.6% at 30 days in patients with advanced cancer (20). Our study had a statistically significantly larger sample size, and we were able to look at factors associated with EM in a multivariable model. Looking at data from randomized controlled trials, a pooled analysis of 18 randomized controlled trials showed that the EM at 60 days was 6.1% and 9.7% in patients treated with ICI with melanoma and lung cancer, respectively. Furthermore, it showed that EM at 60 days was 5% in patients with lung cancer treated with ICI combination (21). The proportion of patients who died early following initiation of ICI was numerically higher in our real-world population compared with outcomes seen in randomized trials of IO. This may be because patients on clinical trials are highly selected and have better performance status and less comorbidities compared with patients treated in routine practice as previously described for cytotoxic chemotherapy (22).
We identified several patient-, tumor-, and system-level factors associated with EM, such as age, tumor site, baseline laboratory values, ESAS score, and type of facility where treatment was given. These factors are similar to those identified in studies assessing EM post chemotherapy and radiation therapy. In a study conducted among patients with colorectal cancer, the Khorana score (site of cancer, hemoglobin, WBC count, platelet count, and BMI) was associated with EM along with older age and worse performance status (11). Similarly in another study assessing EM in patients receiving palliative radiotherapy, the same laboratory values were associated with EM in addition to NLR, urea, albumin, attendance to ED and lung cancer site (9). In addition, in another recent report, factors such age, performance status, NLR, albumin, and Lactate dehydrogenase (LDH) were associated with EM (20). Some of the laboratory values identified were shown to be prognostic in patients with cancer: lower hemoglobin was associated with poorer outcomes in patients with cancer (23-25), high WBC, and platelets. Known markers of inflammation were also shown to be associated with worse survival in cancer studies (26-28). Although higher BMI is known to be a worse prognostic factor for patients with cancer (29), this does not seem to apply in patients treated with ICI. This is supported by recent data showing that excessive adipose tissue is associated with low-grade inflammation, ultimately improving survival in patients treated with ICI (30,31).
Our study has some limitations related to its retrospective nature and risk of selection bias. Furthermore, some important variables such as line of therapy and cause of death were not available to be included in the model, and it would have been valuable to know whether the high rates of EM were higher in heavily pretreated (on second line therapy and beyond) patients as we would expect or not and whether patients with EM died because of disease progression or treatment toxicity. In that setting, we explored the use of steroids, including a high dose of steroids, and found that most patients did not receive a high dose of steroids; therefore, cancer progression may be the main reason of death. However, this needs to be cautiously interpreted due to the data availability for this variable. Also, in this study, we saw a difference in baseline characteristics between patients with different tumor sites, which makes the interpretation of the variable tumor site as a variable associated with EM difficult. As such, although patients with lung cancer had a higher likelihood of EM compared with patients with melanoma in an unadjusted analysis, the results changed in the adjusted model. A potential explanation is that patients with lung cancer were older with more comorbidities and abnormal laboratory values compared with patients with melanoma.
To our knowledge, ours is one of the largest population-level studies to date to estimate EM rate following initiation of ICI, and we were able to identify several factors associated with this negative outcome that could be used to develop prediction tools that can facilitate better patient selection and avoid using ICI in high-risk patients unlikely to benefit that could then be validated in prospective trials.
Our study supports EM as an important clinical outcome that needs to be considered in patients eligible for ICI. The issue of early crossover of survival curves with higher EM seen in patients treated with ICI compared with traditional systemic therapy was previously raised and discussed in the literature (14) but remains poorly understood. Possible explanations are early toxicity, delay in response to IO resulting in disease progression, and primary resistance particularly in patients receiving single-agent ICI. If EM is related to disease progression or primary resistance, combining ICI with other systemic agents may overcome these phenomena and improve outcomes (32).
Mortality early post initiation of ICI is common in patients with cancer treated with ICI. Several patient-, tumor-, and system-level factors are associated with EM and need to be prospectively validated to ultimately develop a predictive tool to use in routine practice for better patient selection and treatment guidance.
Supplementary Material
Acknowledgements
This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The study was completed at the ICES Western site, where core funding is provided by the Academic Medical Organization of Southwestern Ontario, the Schulich School of Medicine and Dentistry, Western University, and the Lawson Health Research Institute. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.
Contributor Information
Jacques Raphael, Division of Medical Oncology, Department of Oncology, London Regional Cancer Program, University of Western Ontario, London, ON, Canada; ICES Western, London, ON, Canada.
Lucie Richard, ICES Western, London, ON, Canada.
Melody Lam, ICES Western, London, ON, Canada.
Phillip Blanchette, Division of Medical Oncology, Department of Oncology, London Regional Cancer Program, University of Western Ontario, London, ON, Canada; ICES Western, London, ON, Canada.
Natasha B Leighl, Division of Medical Oncology and Hematology, Department of Medicine, Princess Margaret Cancer Centre, Toronto, ON, Canada.
George Rodrigues, Division of Radiation Oncology, Department of Oncology, London Regional Cancer Program, University of Western Ontario, London, ON, Canada.
Maureen Trudeau, Division of Medical Oncology, Department of Medicine, Sunnybrook Odette Cancer Centre, Toronto, ON, Canada; Department of Medicine, University of Toronto, Toronto, ON, Canada.
Monika K Krzyzanowska, Division of Medical Oncology and Hematology, Department of Medicine, Princess Margaret Cancer Centre, Toronto, ON, Canada; Department of Medicine, University of Toronto, Toronto, ON, Canada.
Data availability
The dataset from this study is securely held in coded form at ICES. Although data-sharing agreements prohibit ICES from making the dataset publicly available, access can be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS.The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the programs may rely on coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.
Author contributions
Jacques Raphael, MD, MSc (Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Visualization; Writing—original draft; Writing—review and editing), Lucie Richard, MA (Formal analysis; Methodology; Validation; Writing—review and editing), Melody Lam, MSc (Data curation; Formal analysis; Methodology; Writing—review and editing), Phillip Blanchette, MD, MSc (Investigation; Methodology; Writing—review and editing), Natasha B. Leighl, MD, MMSc (Supervision; Validation; Writing—review and editing), George Rodrigues, MD, PhD (Writing—review and editing), Maureen Trudeau, MD, MA (Supervision; Validation; Writing—review and editing), Monika K. Krzyzanowska, MD, MPH (Conceptualization; Project administration; Supervision; Validation; Writing—original draft; Writing—review and editing).
Funding
Medical Oncology Research Fund (MORF) award, London Regional Cancer Program. Grant used to conduct the project and analyses at ICES.
Conflicts of interest
None to declare.
References
- 1. Reck M, Rodríguez-Abreu D, Robinson AG, et al. ; KEYNOTE-024 Investigators. Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer. N Engl J Med. 2016;375(19):1823-1833. [DOI] [PubMed] [Google Scholar]
- 2. Hodi FS, O'Day SJ, McDermott DF, et al. Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med. 2010;363(8):711-723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Motzer RJ, Tannir NM, McDermott DF, et al. ; CheckMate 214 Investigators. Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma. N Engl J Med. 2018;378(14):1277-1290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Carbone DP, Reck M, Paz-Ares L, et al. ; CheckMate 026 Investigators. First-line nivolumab in stage IV or recurrent non-small-cell lung cancer. N Engl J Med. 2017;376(25):2415-2426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Bellmunt J, de Wit R, Vaughn DJ, et al. ; KEYNOTE-045 Investigators. Pembrolizumab as second-line therapy for advanced urothelial carcinoma. N Engl J Med. 2017;376(11):1015-1026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ferris RL, Blumenschein G Jr, Fayette J, et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med. 2016;375(19):1856-1867. doi: 10.1056/NEJMoa1602252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Antonia SJ, Villegas A, Daniel D, et al. ; PACIFIC Investigators. Overall survival with durvalumab after chemoradiotherapy in stage III NSCLC. N Engl J Med. 2018;379(24):2342-2350. doi: 10.1056/NEJMoa1809697. [DOI] [PubMed] [Google Scholar]
- 8. Havel JJ, Chowell D, Chan TA. The evolving landscape of biomarkers for checkpoint inhibitor immunotherapy. Nat Rev Cancer. 2019;19(3):133-150. doi: 10.1038/s41568-019-0116-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Lee SF, Luk H, Wong A, Ng CK, Wong FCS, Luque-Fernandez MA. Prediction model for short-term mortality after palliative radiotherapy for patients having advanced cancer: a cohort study from routine electronic medical data. Sci Rep. 2020;10(1):5779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Boulahssass R, Gonfrier S, Ferrero JM, et al. Predicting early death in older adults with cancer. Eur J Cancer. 2018;100:65-74. [DOI] [PubMed] [Google Scholar]
- 11. Sohal DPS, Kuderer NM, Shepherd FA, et al. Clinical predictors of early mortality in colorectal cancer patients undergoing chemotherapy: results from a global prospective cohort study. JNCI Cancer Spectr. 2017;1(1):pkx009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Soubeyran P, Fonck M, Blanc-Bisson C, et al. Predictors of early death risk in older patients treated with first-line chemotherapy for cancer. J Clin Oncol. 2012;30(15):1829-1834. [DOI] [PubMed] [Google Scholar]
- 13. Anderson J, Handa S, Petrone G, et al. Palliative care utilization and mortality in patients who received inpatient chemotherapy. J Clin Oncol. 2022;40(suppl 28):198. doi: 10.1200/JClinOncol.2022.40.28_suppl.198. [DOI] [Google Scholar]
- 14. Gomes F, Wong M, Battisti NML, et al. Immunotherapy in older patients with non-small cell lung cancer: young International Society of Geriatric Oncology position paper. Br J Cancer. 2020;123(6):874-884. doi: 10.1038/s41416-020-0986-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Winquist E, Kuruvilla S, Nichols AC, et al. Early mortality with immune checkpoint inhibitors (IOs) in solid tumors: an inconvenient truth? J Clin Oncol. 2018;36(suppl 15):12121. [Google Scholar]
- 16. Benchimol EI, Smeeth L, Guttmann A, et al. ; RECORD Working Committee. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12(10):e1001885. doi: 10.1371/journal.pmed.1001885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Vandenberg T, Coakley N, Nayler J, et al. A framework for the organization and delivery of systemic treatment. Curr Oncol. 2009;16(1):4-15. doi: 10.3747/co.v16i1.297. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. O'Brien ME, Borthwick A, Rigg A, et al. Mortality within 30 days of chemotherapy: a clinical governance benchmarking issue for oncology patients. Br J Cancer. 2006;95(12):1632-1636. doi: 10.1038/sj.bjc.6603498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Gibson A, Meyers D, Dean M, Stukalin I, Morris D, Bebb DG. Early mortality associated factors with immune checkpoint inhibition in real-world Canadian NSCLC patients. J Thorac Oncol. 2021;16(3):S348. [Google Scholar]
- 20. Mandaliya HA, Kim S, Quah GT, et al. Mortality within 30 days of immunotherapy (checkpoint inhibitors) in metastatic cancer patients treated at Australian tertiary cancer center. J Clin Oncol. 2019;37(15_suppl):6600-6600. [Google Scholar]
- 21. Mulkey F, By K, Theoret MR, Maher VE, Pazdur R, Sridhara R. Analysis of early mortality in randomized clinical trials evaluating anti-PD-1/PD-L1 antibodies: a systematic analysis by the United States Food and Drug Administration (FDA). J Clin Oncol. 2019;37(suppl 15):2516-2516. [Google Scholar]
- 22. Unger JM, Barlow WE, Martin DP, et al. Comparison of survival outcomes among cancer patients treated in and out of clinical trials. J Natl Cancer Inst. 2014;106(3):dju002. doi: 10.1093/jnci/dju002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Caro JJ, Salas M, Ward A, Goss G. Anemia as an independent prognostic factor for survival in patients with cancer: a systemic, quantitative review. Cancer. 2001;91(12):2214-2221. [PubMed] [Google Scholar]
- 24. Di Maio M, Pisano C, Tambaro R, et al. The prognostic role of prechemotherapy hemoglobin level in patients with ovarian cancer. Front Biosci. 2006;11:1585-1590. [DOI] [PubMed] [Google Scholar]
- 25. Beer TM, Tangen CM, Bland LB, et al. Prognostic value of anemia in newly diagnosed metastatic prostate cancer: a multivariate analysis of southwest oncology group study 8894. J Urol. 2004;172(6 Pt 1):2213-2217. [DOI] [PubMed] [Google Scholar]
- 26. Giessen C, Graeven U, Laubender RP, et al. Prognostic factors for 60-day mortality in first-line treatment of metastatic colorectal cancer (mCRC): individual patient analysis of four randomised, controlled trials by the AIO colorectal cancer study group. Ann Oncol. 2013;24(12):3051-3055. [DOI] [PubMed] [Google Scholar]
- 27. Kohne CH, Cunningham D, Di Costanzo F, et al. Clinical determinants of survival in patients with 5-fluorouracil-based treatment for metastatic colorectal cancer: results of a multivariate analysis of 3825 patients. Ann Oncol. 2002;13(2):308-317. [DOI] [PubMed] [Google Scholar]
- 28. Taucher S, Salat A, Gnant M, et al. Impact of pretreatment thrombocytosis on survival in primary breast cancer. Thromb Haemost. 2003;89(6):1098-1106. [PubMed] [Google Scholar]
- 29. Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med. 2003;348(17):1625-1638. doi: 10.1056/NEJMoa021423. [DOI] [PubMed] [Google Scholar]
- 30. McQuade JL, Daniel CR, Hess KR, et al. Association of body-mass index and outcomes in patients with metastatic melanoma treated with targeted therapy, immunotherapy, or chemotherapy: a retrospective, multicohort analysis. Lancet Oncol. 2018;19(3):310-322. doi: 10.1016/S1470-2045(18)30078-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Yoo S, Chowell D, Valero C, Morris LGT, Chan TA. Outcomes among patients with or without obesity and with cancer following treatment with immune checkpoint blockade. JAMA Netw Open. 2022;5(2):e220448. doi: 10.1001/jamanetworkopen.2022.0448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Zhu S, Zhang T, Zheng L, et al. Combination strategies to maximize the benefits of cancer immunotherapy. J Hematol Oncol. 2021;14(1):156. doi: 10.1186/s13045-021-01164-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The dataset from this study is securely held in coded form at ICES. Although data-sharing agreements prohibit ICES from making the dataset publicly available, access can be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS.The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the programs may rely on coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.