Unexplained increases in serum carcinoembryonic antigen levels in colorectal cancer patients during the postoperative follow-up period: an analysis of its incidence and longitudinal pattern

Figures & data

Figure 1. Study flow chart.

Table 1. Comparative profile of patients according to recurrence.

Variable		No recurrence	Recurrence	p-Value
Number of patients		446	221
Sex	Male	134 (30%)	64 (29%)	0.77
	Female	312 (70%)	157 (71%)
Age, year	Mean ± SD	63.1 ± 12.0	62.8 ± 13.4	0.74
Body mass index	Mean ± SD	22.3 ± 3.4	22.0 ± 3.8	0.28
ECOG PS	0	425 (95%)	204 (92%)	0.35
	1	19 (4%)	16 (7%)
	2	3 (1%)	1 (1%)
Preoperative CEA, ng/ml	Elevated	96 (22%)	131 (59%)	<0.001
Primary tumor location	Colon	255 (57%)	118 (53%)	0.35
	Rectum	191 (43%)	103 (47%)
Preoperative treatment	None	399 (90%)	187 (85%)	0.006
	Chemotherapy	2 (0%)	9 (4%)
	Radiation/chemoradiation therapy	45 (10%)	25 (11%)
Stage	0/I	173 (39%)	14 (6%)	<0.001
	II	133 (30%)	36 (16%)
	III	133 (30%)	103 (47%)
	IV	7 (1%)	68 (31%)
Postoperative chemotherapy	None	311 (70%)	93 (42%)	<0.001
	Fluoropyrimidine analogue	100 (22%)	90 (41%)
	Oxaliplatin-based	35 (8%)	38 (17%)

SD: standard deviation; ECOG PS: Eastern Cooperative Oncology Group Performance Status; CEA: carcinoembryonic antigen.

Table 2. Profile of patients without recurrence according to postoperative CEA elevation.

Variable			No postoperative CEA elevation	Postoperative CEA elevation	p-Value
Number of patients			356	90
Sex		Male	99 (28%)	35 (39%)	0.044
		Female	257 (72%)	55 (61%)
Age, year		Mean ± SD	61.6 ± 11.8	69.1 ± 11.3	<0.001
Age group, years		–49	57 (15%)	7 (9%)	<0.001
		50–59	85 (24%)	9 (10%)
		60–69	122 (34%)	25 (28%)
		70–79	73 (22%)	31 (33%)
		80–	19 (5%)	18 (20%)
Body mass index		Mean ± SD	22.3 ± 3.4	22.5 ± 3.6	0.58
ECOG PS		0	341 (96%)	84 (93%)	0.54
		1	12 (3%)	6 (7%)
		2	3 (1%)	0 (0%)
Preoperative CEA, ng/ml		Elevated	53 (15%)	43 (48%)	<0.001
Preoperative CA 19-9, U/ml		Elevated	29 (8%)	19 (21%)	<0.001
Preoperative AST, U/l		Median (IQR)	19 (16–23)	20 (17–24)	0.18
Preoperative ALT, U/l (men)		Median (IQR)	18 (13–22)	18 (11–25)	0.70
Preoperative ALT, U/l (women)		Median (IQR)	14 (10–19)	14 (11–20)	0.35
Preoperative creatinine, mg/dl (men)	Median (IQR)		0.84 (0.74–0.92)	0.81 (0.71–0.91)	0.52
Preoperative creatinine, mg/dl (women)		Median (IQR)	0.60 (0.54–0.67)	0.60 (0.53–0.69)	0.74
Preoperative CRP, mg/dl		Median (IQR)	0.07 (0.03–0.27)	0.08 (0.03–0.30)	0.54
Primary tumor location		Colon	197 (55%)	58 (64%)	0.12
		Rectum	159 (45%)	32 (36%)
Preoperative treatment		None	321 (90%)	78 (87%)	0.64
		Chemotherapy	2 (1%)	0 (0%)
		Radiation/chemoradiation therapy	33 (9%)	12 (13%)
Stage		0/I	148 (42%)	25 (28%)	0.19
		II	101 (28%)	32 (36%)
		III	102 (29%)	31 (34%)
		IV	5 (1%)	2 (2%)
Postoperative chemotherapy		None	252 (71%)	59 (66%)	0.40
		Fluoropyrimidine analogue	79 (22%)	21 (23%)
		Oxaliplatin-based	25 (7%)	10 (11%)
Whole follow-up period, days		Mean ± SD	3341 ± 1388	2985 ± 1055	0.024
Last laboratory test in the analysis, postoperative days		Mean ± SD	1535 ± 313	1564 ± 281	0.42
Number of postoperative testings		Mean ± SD	17.8 ± 7.6	20.9 ± 8.3	0.001

SD: standard deviation; ECOG PS: Eastern Cooperative Oncology Group Performance Status; CEA: carcinoembryonic antigen; CA: carbohydrate antigen; AST: aspartate aminotransferase; IQR: interquartile range; ALT: alanine aminotransferase; CRP: C-reactive protein.

Figure 2. Evaluation of candidate measures using CEA values for discriminating recurrence from non-recurrence. (A) Distribution of raw CEA levels during the postoperative period. The number of readings for each CEA range was shown according to recurrence. One patient who did not take a CEA test at the time of recurrence was excluded. White bar indicates non-recurrence, and black bar indicates recurrence. (B) Receiver operating characteristic (ROC) curve for the relationship between postoperative CEA value and recurrence. AUC: area under the curve. (C) Distribution of postoperative CEA ratio to the preoperative value. The number of readings for each range of the relative ratio was shown according to recurrence. Patients who did not take a CEA test before colorectal surgery were excluded. White bar indicates non-recurrence, and black bar indicates recurrence. (D) ROC curve for the relationship between postoperative CEA ratio to the preoperative value and recurrence. (E) Distribution of postoperative CEA ratio to the nadir. The number of readings for each range of the relative ratio was shown according to recurrence. One patient who did not take a CEA test at the time of recurrence was excluded. White bar indicates non-recurrence, and black bar indicates recurrence. (F) ROC curve for the relationship between postoperative CEA ratio to the nadir and recurrence.

Figure 3. Typical patterns of false-positive elevations of CEA after colorectal cancer surgery. Examples of longitudinal changes in CEA are shown. The relationships between changes in CEA and seasons are tabulated. The relationships between the patterns of CEA elevations and preoperative CEA levels are also shown.

Figure 4. Scatter plot analyses of serum CEA and other blood test data in colorectal cancer patients without recurrence. Serum CEA was plotted against (A) CA 19-9, (B) AST, (C) ALT in men, (D) ALT in women, (E) creatinine in men, (F) creatinine in women, and (G) CRP. The coefficient (r) and total number of blood tests (n) are shown in each graph. There are some outlying data that were not displayed in the graphs.

Data availability statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.