Comorbidity, not age, is prognostic in patients with advanced pancreatic cancer receiving gemcitabine-based chemotherapy

1. Introduction 

The number of patients with pancreatic cancer has been increasing; pancreatic cancer is now the fourth leading cause of cancer-related death in the United States [1] and the fifth in Japan. Pancreatic cancer often affects elderly patients; the median age at diagnosis is 72 years old, and 42.2% of cases are 75 years or older in the United States [2]. A growing population of these elderly patients has more comorbidities and impaired organ function due to generally increased life expectancy. In systematic reviews, the use of chemotherapy has been shown to improve survival in advanced pancreatic cancer [3]. However, little evidence is available about the benefits and toxicity of chemotherapy in elderly patients, because these patients are typically underrepresented in clinical trials [4]. In clinical practice, less aggressive treatment options or ‘best supportive care’ (BSC) alone is often offered to elderly patients because of fear of increased toxicity.

Gemcitabine has been the standard chemotherapy against advanced pancreatic cancer for more than 10 years, since Burris et al. showed prolonged survival and improved clinical benefit response [5]. There have been two reports of gemcitabine-containing regimens in elderly patients with pancreatic cancer [6], [7]. However, these reports included only selected, fit patients with good performance status (PS) and few comorbidities; thus, generalization of these results may be difficult in the typical clinical setting with unfit elderly patients.

Although chronological age is associated with age-related impairment in organ function, physiological age, which also takes into account comorbidities, may be more important clinically. The Charlson comorbidity index (CCI) [8], [9] is one scale for measuring the severity of comorbidities and has been shown to be reliable in an oncology setting [10]. The aim of this retrospective study was to clarify the impact of age and comorbidity, indicated by CCI, on prognosis in patients with advanced pancreatic cancer.

2. Patients and methods 

2.1. Study design and patients 

All consecutive patients diagnosed with advanced pancreatic cancer at the University of Tokyo Hospital between April 2001 and March 2009 were included. This study was approved by the ethical committee at the University of Tokyo Hospital. Written informed consent was obtained from all patients.

Patient characteristics, including age at diagnosis, gender, performance status (PS), comorbidities, disease stage, treatment protocol, and treatment outcomes, were studied retrospectively. Data on treatment outcomes included the two-course completion rate, dose intensity, adverse effects, objective response, progression-free survival (PFS), and overall survival (OS). Patients were divided into two groups by age: the non-elderly group (<75 years) and the elderly group (≥75 years). We used the Charlson comorbidity index (CCI) for the analysis of patient comorbidity. CCI scores were calculated by the method previously described by Charlson et al. [8]. Comorbid conditions are weighted and scored as shown in Table 1, and CCI score is the sum of these comorbid scores. Patients were divided into three groups by CCI: 0, 1, or ≥2.

Table 1. Charlson comorbidity index.

	Assigned weights	Comorbid conditions
	6	Metastatic solid tumor
		Acquired immunodeficiency syndrome
	3	Moderate or severe liver disease
	2	Hemiplegia
		Moderate or severe renal disease
		Diabetes with end organ damage
		Any tumor
		Leukemia
		Lymphoma
	1	Myocardial infarct
		Congestive heart failure
		Peripheral vascular disease
		Cerebrovascular disease
		Dementia
		Chronic pulmonary disease
		Connective tissue disease
		Ulcer disease
		Mild liver disease
		Diabetes with end organ damage

3. Results 

3.1. Patient characteristics 

Between April 2001 and March 2009, 237 patients were diagnosed with unresectable pancreatic cancer at the University of Tokyo Hospital. Baseline characteristics are shown in Table 2. The overall introduction rate of chemotherapy was 77.2%. Reasons for not receiving chemotherapy were impaired general condition due to advanced pancreatic cancer in 26 patients, patient refusal in 17, and comorbidities in 11, including six patients with dementia.

Table 2. Characteristics of all patients, grouped by age and comorbidity.

			Overall	Group by age			Group by CCI
				Non-elderly	Elderly	p-Value	CCI 0	CCI 1	CCI ≥2	p-Value
	Number of patients		237	168 (70.9%)	69 (29.1%)		69 (29.1%)	98 (41.4%)	70 (29.5%)
	Age at diagnosis	Mean±SD	67.8±10.7	62.8±8.2	80.1±4.3	<0.001	64.9±11.5	66.9±10.3	71.9±9.4	<0.001
	Group by age	Non-elderly	168 (70.9%)				69.6%	80.6%	58.6%	0.008
		Elderly	69 (29.1%)				30.4%	19.4%	41.4%
	Gender	Male	142 (59.9%)	61.9%	55.1%	0.382	52.2%	60.2%	67.1%	0.206
		Female	95 (40.1%)	38.1%	44.9%		47.8%	39.8%	32.9%
	PS	0	83 (35.0%)	38.1%	27.5%	0.002	37.7%	37.8%	28.6%	0.314
		1	100 (42.2%)	45.8%	33.3%		47.8%	37.8%	42.9%
		2	38 (16.0%)	11.9%	26.1%		8.7%	16.3%	22.9%
		3	16 (6.8%)	4.2%	13.0%		5.8%	8.2%	5.7%
	CCI	Mean±SD	1.24±1.24	1.17±1.19	1.41±1.34	0.200	0	1	2.81±1.05	<0.001
	Group by CCI	0	69 (29.1%)	28.6%	30.4%	0.008
		1	98 (41.4%)	47.0%	27.5%
		≥2	70 (29.5%)	24.4%	42.0%
	Tumor location	Head	120 (50.6%)	47.0%	59.4%	0.088	55.1%	49.0%	48.6%	0.692
		Body or tail	117 (49.4%)	53.0%	40.6%		44.9%	51.0%	51.4%
	Distant metastasis	Yes	148 (62.4%)	64.9%	56.5%	0.240	73.9%	59.2%	55.7%	0.057
		No	89 (37.6%)	35.1%	43.5%		26.1%	40.8%	44.3%
	Sites of distant metastasis	Liver	108 (45.6%)	47.6%	40.6%	0.389	47.8%	45.9%	42.9%	0.855
		Lymph node	107 (45.1%)	50.6%	31.9%	0.010	46.4%	50.0%	37.1%	0.254
		Lung	30 (12.7%)	10.7%	17.4%	0.197	15.9%	14.3%	7.1%	0.253
		Peritoneum	60 (25.3%)	23.2%	30.4%	0.254	27.5%	25.5%	22.9%	0.815
	CA19-9 (U/mL)	Median (Range)	416.5 (1–3,270,000)	440 (1–534,000)	285.5 (1–3,270,000)	0.513	658 (1–534,000)	368 (1–3,270,000)	485.5 (1–145,600)	0.534
	Treatment	Chemotherapy	183 (77.2%)	84.5%	59.4%	<0.001	78.3%	81.6%	70.0%	0.209
		BSC	54 (22.8%)	15.5%	40.6%		21.7%	18.4%	30.0%
	OS (months)	All	9.2 (8.4–10.4)	9.5 (8.6–11.2)	8.3 (5.0–10.6)	0.711	10.2 (7.7–13.3)	9.5 (7.2–11.3)	8.3 (6.0–10.4)	0.356
		BSC alone	2.8 (2.1–3.6)	2.4 (0.9–3.6)	3.4 (1.0–6.2)	0.145	2.5 (0.7–8.6)	3.5 (1.1–4.7)	2.2 (0.7–3.3)	0.337
		Chemotherapy	10.8 (9.5–13.2)	10.8 (9.4–13.2)	10.9 (7.0–18.4)	0.489	11.7 (9.4–17.2)	10.5 (8.8–12.9)	10.2 (8.4–13.2)	0.485

Abbreviations: PS, performance status; CCI, Charlson comorbidity index; GEM, gemcitabine monotherapy; GEM+S-1, gemcitabine and S-1 combination therapy; OS, overall survival; BSC: best supportive care.

The mean age at diagnosis was 67.8 years, and 69 patients (29.1%) were elderly (75 years or older). Elderly patients had poorer PS (p=0.002) and a lower introduction rate of chemotherapy (59.4% in the elderly group vs. 84.5% in the non-elderly group; p<0.001). Regarding comorbidities, CCI was 0 in 29.1%, 1 in 41.4%, and ≥2 in 29.5%. The elderly group correlated with CCI grouping (p=0.008). Grouping by CCI was not statistically significantly associated with PS (p=0.314) or chemotherapy use (p=0.209).

OS was 9.2 months (95% CI, 8.4–10.4 months) in all 237 patients, 2.8 months (95% CI, 2.1–3.6 months) in patients with BSC alone, and 10.8 months (95% CI, 9.5–13.2 months) in patients receiving chemotherapy (p<0.001).

3.2. Chemotherapy regimens, dose intensity, and safety 

Chemotherapy regimens, dose intensity, and adverse effects by age or comorbidity are shown in Table 3. Administered chemotherapy regimens were gemcitabine monotherapy in 76.5% and gemcitabine and S-1 combination therapy in 23.5%. Groups by age or comorbidity were not associated with the regimens. Initial dose of gemcitabine was reduced in 10 patients who received gemcitabine monotherapy. The reasons of initial dose reduction were age (N=2) and comorbidity (N=8). The rate of initial dose reduction in gemcitabine monotherapy was 12.9% in the elderly group and 5.5% in the non-elderly group (p=0.228), and it was 2.7% in the CCI 0 group, 1.6% in the CCI 1 group, and 20.5% in the CCI ≥2 group (p=0.001). On the other hand, the two-cycle completion rate was lower in the elderly group (73.2%) than the non-elderly group (89.4%; p=0.019), but was not statistically significantly different by comorbidity, although the CCI 0 group tended to have a higher completion rate. The RDI of gemcitabine monotherapy was lower in the elderly group (p=0.001), whereas the RDI of the gemcitabine and S-1 combination therapy was lower in the comorbidity group (p=0.090 for gemcitabine and p=0.016 for S-1).

Table 3. Clinical outcomes, dose intensity, and adverse effects by age and comorbidity.

			Group by age			Group by CCI
			Non-elderly	Elderly	p-Value	CCI 0	CCI 1	CCI ≥2	p-Value
	Number of patients		142	41		54	80	49
	Protocol	GEM	76.8%	75.6%	0.838	68.5%	80.0%	79.6%	0.267
		GEM+S-1	23.2%	24.4%		31.5%	20.0%	20.4%
	Two-cycle completion rate		89.4%	73.2%	0.019	90.7%	82.5%	85.7%	0.580
	ORR		5.6%	17.1%	0.046	13.0%	5.0%	8.2%	0.242
	PFS (months)		5.1	4.7	0.341	4.5	5.3	5.3	0.277
	OS (months)		10.8	10.9	0.489	11.7	10.5	10.2	0.485
	RDI of each protocol
	GEM	GEM	68.0±21.0%	52.7±28.5%	0.001	68.4±21.8%	66.4±22.7%	58.3±25.9%	0.133
	GEM+S-1	GEM	91.2±18.9%	83.9±24.9%	0.909	98.4±4.3%	84.6±4.4%	90.3±5.4%	0.090
		S-1	92.0±13.5%	73.9±21.3%	0.279	94.6±5.6%	76.1±5.8%	69.6±7.1%	0.016
	Adverse effects
	Hematological
	Neutropenia	All grades	62.7%	53.7%	0.365	63.0%	57.5%	63.3%	0.777
		Grades 3–4	31.7%	34.1%	0.850	42.6%	25.0%	32.7%	0.106
	Anemia	All grades	78.9%	75.6%	0.671	88.9%	70.0%	79.6%	0.032
		Grades 3–4	3.5%	12.2%	0.047	7.4%	1.3%	10.2%	0.050
	Thrombocytopenia	All grades	59.9%	63.4%	0.720	57.4%	53.8%	75.5%	0.040
		Grades 3–4	2.1%	4.9%	0.312	3.7%	0	6.1%	0.077
	Non-hematological
	Anorexia	All grades	44.4%	34.1%	0.284	51.9%	42.5%	30.6%	0.095
		Grades 3–4	0.7%	2.4%	0.399	0	1.3%	2.0%	0.741
	Nausea	All grades	33.8%	22.0%	0.182	44.4%	30.0%	18.4%	0.016
		Grades 3–4	0.7%	1.4%	0.399	0	1.3%	2.0%	0.741
	Vomiting	All grades	9.2%	12.2%	0.558	13.0%	10.0%	6.1%	0.513
		Grades 3–4	1.4%	0	1.000	0	2.5%	0	0.505
	Diarrhea	All grades	8.5%	12.2%	0.541	11.1%	12.5%	2.0%	0.105
		Grades 3–4	0.7%	0	1.000	0	1.3%	0	1.000
	Constipation	All grades	24.6%	29.3%	0.548	16.7%	27.5%	32.7%	0.149
		Grades 3–4	0	0		0	0	0
	Stomatitis	All grades	3.5%	9.8%	0.115	5.6%	3.8%	6.1%	0.753
		Grades 3–4	0	0		0	0	0
	Skin rash	All grades	10.6%	7.3%	0.767	11.1%	12.5%	4.1%	0.272
		Grades 3–4	2.8%	2.4%	1.000	1.9%	5.0%	0	0.329
	Hyperpigmentation	All grades	7.0%	7.3%	1.000	13.0%	5.0%	4.1%	0.160
		Grades 3–4	0	0		0	0	0

Abbreviations: ORR, objective response; RDI, relative dose intensity; GEM, gemcitabine monotherapy; GEM+S-1, gemcitabine and S-1 combination therapy; CCI, Charlson comorbidity index.

Adverse effects did not differ markedly between groups by age or comorbidity. Regarding age, only grade 3–4 anemia was more commonly observed in the elderly group. Regarding comorbidity, all grade anemia and nausea were more often observed in the no comorbidity group, whereas all grade thrombocytopenia was more frequently observed in patients with CCI ≥2.

3.3. Efficacy of chemotherapy 

Efficacy of chemotherapy was compared between groups by age or comorbidity (Table 3). ORR by RECIST was 5.6% in the non-elderly group and 17.1% in the elderly group (p=0.046), and it was 13.0% in the CCI 0 group, 5.0% in CCI 1, and 8.2% in the CCI ≥2 group (p=0.242). PFS was not significantly affected by age or comorbidity: 5.1 months in the non-elderly group versus 4.7 months in the elderly group (p=0.341), and 4.5 months in the CCI 0 group versus 5.3 months in both the CCI 1 and CCI ≥2 groups (p=0.277). OS with chemotherapy was similar between age groups (Fig. 1): 10.8 months in the non-elderly group and 10.9 months in the elderly group (p=0.489). Grouping by CCI (Fig. 2) was not significantly correlated with survival (p=0.485), but the OS of 11.7 months in the CCI 0 group was slightly longer than in the CCI 1 (10.5 months) or CCI ≥2 group (10.2 months).

View Large Image Download to PowerPoint

Fig. 1. Kaplan–Meier curves for survival by age in patients receiving chemotherapy. No significant difference (p=0.489) was found between the non-elderly (10.8 months, solid line) and elderly groups (10.9 months, broken line).

View Large Image Download to PowerPoint

Fig. 2. Kaplan–Meier curves for survival by comorbidity in patients receiving chemotherapy. Grouping by CCI was not significantly correlated with survival (p=0.485), but the OS (11.7 months) in the CCI 0 group (solid line) was slightly longer than those in the CCI 1 group (10.5 months, broken line) and the CCI ≥2 group (10.2 months, thin line).

3.4. Univariate and multivariate analyses of survival 

Table 4 shows the results of univariate and multivariate analyses of prognostic factors. PS, CA19-9, distant metastasis, and chemotherapy use were shown to be prognostic by the univariate analysis. Age and comorbidity were not significantly prognostic by univariate analysis, although higher CCI tended to be associated with poorer prognosis (p=0.156). To evaluate the prognostic impact of age and comorbidity, multivariate Cox proportional regression analysis was performed, including age, comorbidity, and the four significant factors from the univariate analysis.

Table 4. Univariate and multivariate analyses of prognostic factors.

	Variables		Univariate		Multivariate
			HR (95% CI)	p-Value	HR (95% CI)	p-Value
	Age	<75	1	0.712	1	0.215
		≥75	1.06 (0.76–1.46)		0.80 (0.56–1.13)
	Gender	Male	1	0.075
		Female	0.77 (0.57–1.03)
	PS	0	1	<0.001	1	<0.001
		≥1	2.30 (1.68–3.19)		1.90 (1.37–2.68)
	CA19-9 (U/mL)	10,000/increase	1.01 (1.01–1.02)	0.002	1.01 (1.00–1.02)	0.009
	Distant metastasis	No	1	0.002	1	<0.001
		Yes	1.61 (1.19–2.19)		1.91 (1.39–2.64)
	Treatment	BSC alone	1	<0.001	1	<0.001
		Chemotherapy	0.26 (0.18–0.38)		0.25 (0.17–0.38)
	CCI	0	1	0.156	1	0.027
		1	1.15 (0.95–1.38)		1.25 (1.03–1.52)
		≥2	1.31 (0.90–1.92)		1.55 (1.05–2.30)

Abbreviations: PS, performance status; CCI, Charlson comorbidity index; BSC, best supportive care.

In this multivariate analysis, CCI was associated with poor prognosis. Compared with CCI 0, the HRs of CCI 1 and ≥2 were 1.25 and 1.55, respectively (p=0.027). PS, CA19-9, metastasis, and chemotherapy use were also independent prognostic factors in the multivariate analysis. In contrast, age was not related to prognosis in either the univariate or multivariate analysis.

4. Discussion 

In this retrospective analysis, gemcitabine-based chemotherapy in elderly patients with advanced pancreatic cancer was safe and effective, as previously reported [6], [7]. Multivariate analysis revealed that comorbidity, indicated by increased CCI, but not age, was a poor prognostic factor, suggesting that comorbidity was more important than chronological age in the treatment of advanced pancreatic cancer.

An increasing proportion of elderly patients with cancer is expected because of the general increase in life expectancy, but treatment of elderly patients remains a challenge for oncologists. Elderly patients are often underrepresented in clinical trials because higher age is often included in the exclusion criteria [4], and evidence is therefore lacking in these patients. Unfortunately, elderly patients are often suboptimally treated for fear of potentially increased toxicity and decreased benefits, although little evidence supports this practice. The efficacy and safety of surgical resection in elderly patients with resectable pancreatic cancer have been reported [13], [14], [15]. However, in a palliative setting such as the use of chemotherapy for advanced pancreatic cancer, the decision between palliative chemotherapy and BSC alone may be more difficult. Chemotherapy in elderly patients has been reported to be effective in advanced non-small-cell lung cancer [16], [17] and colorectal cancer [18], [19], [20]. In elderly patients with advanced pancreatic cancer, few reports discuss palliative chemotherapy [21]. Locher et al. [6] conducted a prospective observational study of fixed-dose rate gemcitabine in 39 patients aged 70 years and older and concluded that gemcitabine was feasible in elderly patients with good PS and without major comorbidity. Marechal et al. [7] reported the tolerability and efficacy of gemcitabine and gemcitabine-based regimens in patients aged 70 years and older, but this retrospective study evaluated only patients with good PS who were eligible for phase 2 and 3 trials.

Unlike these previous reports, unfit patients who received BSC alone were included in our study and the results may be more relevant to clinical practice. In patients receiving chemotherapy, PFS and OS were not significantly different between groups by age. Thus, we consider that chronological age alone is not a contraindication for chemotherapy. Regarding safety, however, caution must be exercised in elderly patients. Although adverse events did not increase with age, except grade 3–4 anemia, the two-course completion rate and RDI with gemcitabine monotherapy was higher in non-elderly patients, showing that more intense chemotherapy was administered in the non-elderly group. The reason that similar OS was achieved in the elderly group in spite of the lower two-cycle completion rate and RDI is unclear. However, we thought that in elderly patients who completed at least two cycles of chemotherapy could obtain survival benefit even in the lower dose intensity. Thus, the appropriate dose intensity in elderly patients with reduced organ function remains to be clarified.

In addition to chronological age, comorbidity is the problem especially encountered when treating elderly patients, as was confirmed by the positive correlation between group and age and between group and comorbidity in our study. Thus, assessment of comorbidity is important with increased age. CCI is a simple index for the assessment of comorbidity developed by Charlson et al. [8]. In elderly patients with colorectal cancer, CCI was reported to be a prognostic factor for poor survival [22], [23]. In non-small-cell lung cancer, too, CCI, rather than age over 65 years, was associated with poorer survival [17]. In our multivariate analysis, CCI was one of the prognostic factors for survival in advanced pancreatic cancer, whereas age was not. CCI and PS were independent prognostic factors in our patient population. Comorbidity was reported to be independent of functional status, defined by PS, in elderly patients with cancer [10], and functional status and comorbidity were independent prognostic factors in patients with lung cancer [17], [24], [25]. Thus, assessment of comorbidity is important in treating patients with advanced pancreatic cancer, and stratification by CCI should be considered in prospective trials, especially in trials including elderly patients.

Why CCI was associated with prognosis is unclear, but considering the fatality of advanced pancreatic cancer, comorbidity itself is unlikely to directly affect survival. It is possible that chemotherapy might not be offered to patients with comorbidities, because in patients with CCI ≥2, the rate of chemotherapy was slightly lower (70.0%) than that in the CCI 0 group (78.3%) or the CCI 1 group (81.6%). Additionally, comorbidities could lead to intolerance or dose reduction of chemotherapy, as the two-cycle completion rate and RDI in the CCI 0 group was somewhat higher than that in the CCI 1 or ≥2 group. Thus, comorbidity might indirectly affect survival both by reluctance of physicians to offer chemotherapy and the tendency to offer less intensive chemotherapy in patients with comorbidities.

This study has some limitations. First, it was a retrospective study at a single institution. Although the inclusion of patients with BSC alone in our study helped the generalization of our results to clinical practice, a referral bias is inevitable because our institution is an academic medical center. Moreover, no clear criteria were established for initial dose modification. In elderly patients with comorbidities, dose modification of gemcitabine monotherapy was determined at the discretion of each physician. Thus, no conclusion as to whether initial dose modification is necessary in elderly patients or in patients with comorbidities can be drawn from our analyses. The results of the adverse effects must also be interpreted with caution; although adverse effects did not differ markedly by age or comorbidity, dose intensity was somewhat reduced in elderly patients and in patients with comorbidities. Thus, the safety of gemcitabine-based chemotherapy should be further examined in a properly designed prospective trial in these patients.

In conclusion, our results showed that gemcitabine-based chemotherapy can be an effective treatment, without significant toxicity, in elderly patients. Comorbidity, assessed by CCI, rather than chronological age, was prognostic, and CCI may be useful in treatment choices in clinical practice in patients with advanced pancreatic cancer.

Conflict of interest 

None of the authors have any financial conflicts of interest to declare.

Reviewer 

Dr. Keiji Hanada, Onomichi General Hospital, Department of Gastroenterology, Onomichi, Hiroshima, Japan.