Patient characteristics
A total of 368 qualified patients were identified for cohort 1 in this study. Of the 368 patients, 105 had readmission data and were selected for cohort 2. There were 201 males (54.62%) and 167 females (45.38%) in cohort 1. Most patients had stage IV disease (92.12%). In the classification of tumors, gastrointestinal tumors (189, 51.36%) accounted for the largest proportion of patients, while head and neck cancer (17, 4.62%) accounted for a relatively small proportion. Most patients received the best supportive care (275, 74.73%). There was no significant difference in the number of patients with ECOG < 3 (216, 58.69%) and ECOG ≥ 3 (152, 41.31%). Patients with poor nutritional status accounted for the vast majority of the study patients (289, 78.53%).
In cohort 2, there was a similar number of males (55, 52.38%) and females (50, 47.62%). Patients with stage IV disease (89, 84.76%) and gastrointestinal tumors (58, 55.24%) accounted for the vast majority of cases. Unlike cohort 1, cohort 2 had more patients who received palliative chemoradiotherapy (PCR) than best supportive care (BSC) (54 [51.43%] vs. 51 [48.57%]). There were far more patients with ECOG < 3 (79, 75.24%) than ECOG ≥ 3 (26, 24.76%). The number of patients with good nutritional status (41, 39.05%) was less than the number with poor nutritional status (64, 60.95%).
Table 1
Comparisons of baseline clinicopathological features based on Hb in cohort 1 (N = 368)
Clinicopathological features
|
N (%)
|
Hb
|
P value
|
HHb (n = 236)
|
LHb (n = 132)
|
Age (Mean ± SD)
|
368 (63.56±12.881)
|
64.42±12.423
|
62.04±13.576
|
0.09
|
Gender
|
|
|
|
0.008
|
Male
|
201(54.62%)
|
141(59.74%)
|
60(45.45%)
|
|
Female
|
167(45.38%)
|
95(40.26%)
|
72(54.55%)
|
|
Tumor stage
|
|
|
|
|
III
|
29(7.88%)
|
20(8.47%)
|
9(6.82%)
|
0.572
|
IV
|
339(92.12%)
|
216(91.53%)
|
123(93.18%)
|
|
Primary tumor site
|
|
|
|
0.709
|
Gastrointestinal tumors
|
189(51.36%)
|
118(50%)
|
71(53.79%)
|
|
Thoracic cancers
|
83(22.55%)
|
56(23.73%)
|
27(20.45%)
|
|
Urogenital neoplasms
|
60(16.31%)
|
36(15.25%)
|
24(18.18%)
|
|
Head and neck cancers
|
17(4.62%)
|
12(5.08%)
|
5(3.79%)
|
|
Other tumors
|
19(5.16%)
|
14(5.94%)
|
5(3.79%)
|
|
Palliative care
|
|
|
|
0.004
|
PCR
|
93(25.27%)
|
71(30.08%)
|
22(16.67%)
|
|
BSC
|
275(74.73%)
|
165(69.92%)
|
110(83.33%)
|
|
Family history
|
|
|
|
0.372
|
No
|
260(70.65%)
|
163(69.07%)
|
97(73.48%)
|
|
Yes
|
108(29.35%)
|
73(30.93%)
|
35(26.52%)
|
|
ECOG
|
|
|
|
0.584
|
< 3
|
216(58.69%)
|
141(59.75%)
|
75(56.82%)
|
|
³ 3
|
152(41.31%)
|
95(40.25%)
|
57(43.18%)
|
|
Comorbidity
|
|
|
|
0.998
|
No
|
223(60.59%)
|
143(60.59%)
|
80(60.61%)
|
|
Yes
|
145(39.41%)
|
93(39.41%)
|
52(39.39%)
|
|
Nutritional status
|
|
|
|
0.481
|
Normal
|
79(21.47%)
|
48(20.34%)
|
31(23.49%)
|
|
Abnormal
|
289(78.53%)
|
188(79.66%)
|
101(76.51%)
|
|
Abbreviations: Hb, hemoglobin; SD, standard deviation; LHb, low Hb (pretreatment NC £ 100); HHb, high Hb (pretreatment NC > 100); BSC, best supportive care; PCR, palliative chemoradiotherapy; ECOG, Eastern Cooperative Oncology Group |
Table 2
Comparisons of baseline clinicopathological features based on Hb in cohort 2 (N = 105)
Clinicopathological features
|
N (%)
|
Hb
|
P value
|
Descending
|
Ascending
|
|
Age (Mean ± SD)
|
105(62.92±12.012)
|
63(62.11±11.278)
|
42(64.14±13.081)
|
0.398
|
Gender
|
|
|
|
0.69
|
Male
|
55(52.38%)
|
34(53.97%)
|
21(50%)
|
|
Female
|
50(47.62%)
|
29(46.03%)
|
21(50%)
|
|
Tumor stage
|
|
|
|
0.438
|
III
|
16(15.24%)
|
11(17.46%)
|
5(11.91%)
|
|
IV
|
89(84.76%)
|
52(82.54%)
|
37(88.09%)
|
|
Primary tumor site
|
|
|
|
0.653
|
Gastrointestinal tumors
|
58(55.24%)
|
33(52.38%)
|
25(59.53%)
|
|
Thoracic cancers
|
14(13.33%)
|
8(12.69%)
|
6(14.28%)
|
|
Urogenital neoplasms
|
24(22.86%)
|
17(26.99%)
|
7(16.67%)
|
|
Head and neck neoplasm
|
6(5.71%)
|
4(6.34%)
|
2(4.76%)
|
|
Other tumors
|
3(2.86%)
|
1(1.59%)
|
2(4.76%)
|
|
Palliative care
|
|
|
|
0.3
|
PCR
|
54(51.43%)
|
35(55.56%)
|
19(45.24%)
|
|
BSC
|
51(48.57%)
|
28(44.44%)
|
23(54.76%)
|
|
Family history
|
|
|
|
0.226
|
No
|
73(69.52%)
|
41(65.08%)
|
32(76.19%)
|
|
Yes
|
32(30.43%)
|
22(34.92%)
|
10(23.81%)
|
|
ECOG
|
|
|
|
0.854
|
< 3
|
79(75.24%)
|
47(74.60%)
|
32(76.19%)
|
|
³ 3
|
26(24.76%)
|
16(25.40%)
|
10(23.81%)
|
|
Comorbidity
|
|
|
|
0.218
|
No
|
65(61.90%)
|
42(66.67%)
|
23(54.76%)
|
|
Yes
|
40(38.10%)
|
21(33.33%)
|
19(45.24%)
|
|
Nutritional status
|
|
|
|
0.288
|
Normal
|
41(39.05%)
|
22(34.92%)
|
19(45.24%)
|
|
Abnormal
|
64(60.95%)
|
41(65.08%)
|
23(54.76%)
|
|
Abbreviations: Hb, hemoglobin; SD, standard Deviation; BSC, best supportive care; PCR, palliative chemoradiotherapy; ECOG, Eastern Cooperative Oncology Group |
Determination of the Hb cut-off value
In the clinical environment, it is common to convert continuous variables to binary variables. In this study, the patients were divided into "high" and "low" groups, which was convenient for diagnostic or prognostic predictions. However, for patients with advanced cancer, the accepted clinical threshold was not available. Using X-tile based on a minimum p-value algorithm[10], the optimal cut-off value for Hb was determined to be 100 g/L. Therefore, the patients with HB levels greater than 100 g/L were assigned to the high HB (HHB) group and the patients with HB levels less than or equal to 100 g/L were part of the low HB (LHB) group.
Association of Hb with clinicopathological features
The relationships between the clinicopathological features and Hb (low vs. high) for cohort 1 are shown in Table 1. There were no differences in age, tumor site, family history, ECOG, nutritional status, or comorbidity between the HHB and LHB groups. In contrast, palliative treatment and gender were significantly associated with Hb. The characteristics of cohort 2 are shown in Table 2, where patients were divided into descending (changes < 0) and ascending (changes > 0) groups. There was no significant difference in the proportion of patients with specific clinicopathological features between the two groups.
Association of Hb with OS
Patients in cohort 1 were divided into HHb and LHb groups. The survival time for the HHB group was significantly longer than that of the LHB group (median survival time: 84 days [95% CI 61–107] vs. 41 days [95% CI 31–51), P < 0.001) (Figure 2). Multivariate Cox proportional hazard model showed that tumor stage, ECOG score, palliative treatment, and Hb were independent prognostic factors for OS. After controlling for important confounding variables, HB was still significantly correlated with OS (HHB vs. LHB, HR 0.733, 95% CI 0.558–0.909) (Table 3).
We evaluated cohort 2 to verify the prognostic significance of dynamic changes in Hb by dividing into ascending and descending groups. We found that patients who changed from the HHb group pretreatment to LHb after the second admission had the worst OS (LHb → LHb median survival time [109, 95% CI 49–169] vs. LHb → HHb median survival time [104, 95% CI 0–276] vs. HHb → LHb median survival time [68, 95% C: 45–90] vs. HHb → HHb median survival time [234, 95% CI 168–300], P = 0.013) (Figure 3). The related subgroups were further analyzed and significant results were obtained after adjusting for demographic and disease-specific factors (Table 4).
Table 3
Multivariate Cox regression analysis for Hb (N = 368)
Prognostic factors
|
Hb Model#
|
Adjusted HR (95% CI)
|
P value
|
Gender (male vs. female)
|
0.834(0.634-1.034)
|
0.09
|
Age
|
0.998(0.988-1.008)
|
0.635
|
Primary tumor site
|
1.008(0.376-1.640)
|
0.156
|
Tumor stage (IV vs. III)
|
7.089(2.710-11.469)
|
<0.001
|
ECOG (< 3 vs. ≥ 3)
|
0.606(0.461-0.751)
|
<0.001
|
Nutritional status (normal vs. abnormal)
|
1.068(0.763-1.373)
|
0.879
|
Palliative care (PCR vs. BSC)
|
0.573(0.401-0.745)
|
<0.001
|
Comorbidity (yes vs. no)
|
0.855(0.640-1.071)
|
0.151
|
Family history (yes vs. no)
|
1.217(0.915-1.520)
|
0.203
|
HHb vs. LHb
|
0.733(0.558-0.909)
|
0.006
|
Abbreviations: Hb, hemoglobin; HR, hazard ratio; CI, confidence interval; LHb, low Hb (pretreatment ≤ 100); HHb, high Hb (pretreatment > 100); BSC, best supportive care; PCR, palliative chemoradiotherapy; ECOG, Eastern Cooperative |
Oncology Group.
#The potential prognostic factors used for adjustment in the Cox regression model controlling for gender; age; primary tumor site; tumor stage; ECOG; nutrient status; palliative care; comorbidity; family history.
Table 4
Adjusted HRs for overall survival stratified by changes in Hb in cohort 2 (N = 105)
Hb
|
Hb Model#
|
Adjusted HR (95% CI)
|
P value
|
Ascending
|
Reference
|
|
Descending
|
0.684(0.366-1.002)
|
0.051
|
LHb → LHb
|
Reference
|
|
LHb → HHb
|
1.841(0.211-3.472)
|
0.828
|
HHb → LHb
|
4.003(1.421-6.585)
|
0.004
|
HHb → HHb
|
0.701(0.293-1.109)
|
0.098
|
Abbreviations: Hb, hemoglobin; HR, hazard ratio; CI, confidence interval; LHb, low Hb (pretreatment £ 100); HHb, high Hb (pretreatment > 100). #The potential prognostic factors used for adjustment in the Cox regression model controlling for gender, age, primary tumor site, tumor stage, ECOG, nutrient status, palliative care, comorbidity, and family history.
|