Ability of lactulose breath test results to accurately identify colorectal polyps through the measurement of small intestine bacterial overgrowth

Table 1 Demographics and comorbidity of the study subjects, n (%)

	Overall (N = 382)	Polyps (n = 169)	Non-polyps (n = 213)	P value
Age, yr	57.0 ± 14.0	62.1 ± 11.7	53.0 ± 14.4	0
19-45	70 (18)	11 (7)	59 (28)	0
46-60	150 (39)	65 (38)	85 (40)	0.428
61-92	162 (43)	93 (55)	69 (32)	0
Male	213 (56)	108 (64)	105 (49)	0.003
Bilestone	34 (9)	18 (11)	16 (8)	0.187
Constipation	67 (18)	38 (22)	29 (14)	0.017
Diabetes	53 (14)	32 (19)	21 (10)	0.008
Fatty liver/cirrhosis	127 (33)	70 (41)	57 (27)	0.002
FGID	37 (10)	9 (5)	28 (13)	0.007
GERD	54 (14)	19 (11)	35 (16)	0.096
Hyperlipidemia	60 (16)	33 (20)	27 (13)	0.046
Hypertension	108 (28)	64 (38)	44 (21)	0
Hyperuricemia	42 (11)	20 (12)	22 (10)	0.38
IBS	49 (13)	14 (8)	35 (16)	0.013
PU	31 (8)	24 (14)	7 (3)	0

Values presented as mean ± SD, or n (%) of observations. P values were from one-side Fisher exact statistics, with bold font for those less than 0.05. FGID: Functional gastrointestinal disorders; GERD: Gastroesophageal reflux disease; IBS: Irritable bowel syndrome; PU: Peptic ulcer.

Table 2 Small intestine bacteria overgrowth distribution between polyp & non-polyp groups

SIBO (+)	Overall (N = 382)	Polyps (n = 169)	Non-polyps (n = 213)	P value
By methane	245 (64)	120 (71)	125 (59)	0.014a
By hydrogen 90 min	119 (31)	70 (41)	49 (23)	0.000a
By combined M and H	279 (73)	136 (80)	143 (67)	0.004a

^aP < 0.05, polyps vs non-polyps.

SIBO: Small intestine bacteria overgrowth.

Table 3 The peak values of methane and hydrogen in inflammatory/hyperplastic polyp, adenomatous polyp and non-polyp groups

Peak values	Polyps (n = 169)	Non-polyps (n = 213)		P value
	Inflammatory/hyperplastic polyp	Adenomatous polyp
Methane	208.2	197.86	182.52	0.168
Hydrogen within 90 min	209.53b	220.87b	172.51	0.001a

^aP < 0.05 was considered to indicate a statistically significant difference between inflammatory/hyperplastic polyp, adenomatous polyp and non-polyp groups.

^bP < 0.05 was considered to indicate a statistically significant difference between inflammatory/hyperplastic polyp and non-polyp groups or between adenomatous polyp and non-polyp groups. Univariate analysis was performed using the nonparametric tests (Kruskal-Wallis independent samples).

Table 4 Performance of prediction models for colorectal polyps

Model No.	Value cutoff (ppm) for subset	N	n (polyp/non-polyp)	Mean ppm (polyp)	Mean ppm (non-polyp)	Mean ppm (P value)	Sensitivity	Specificity	Accuracy
1	Methane peak value (not applied)	382	169/213	12.82	12.06	0.074a	0.427	0.578	0.511
2	Methane peak value (≥ 5)	359	160/199	13.34	12.66	0.084a	0.417	0.571	0.502
3	Methane peak value (≥ 10)	245	120/125	15.28	15.55	0.905	0.443	0.390	0.416
4	Hydrogen peak value (not applied)	380	169/211	52.34	45.48	0.091a	0.408	0.604	0.517
5	Hydrogen peak value (≥ 10)	310	140/170	62.05	55.24	0.090a	0.383	0.590	0.496
6	Hydrogen peak value (≥ 20)	259	125/134	67.85	66.02	0.776	0.400	0.439	0.421
7	Hydrogen rise value (not applied)	372	165/207	41.94	37.40	0.121	0.391	0.582	0.498
8	Hydrogen rise value (≥ 10)	280	130/150	52.12	49.93	0.406	0.380	0.477	0.432
9	Hydrogen rise value (≥ 20)	217	108/109	60.06	63.39	0.479	0.492	0.375	0.433
10	Hydrogen rise value (≥ 20 by 90 min)	119	70/49	47.74	50.10	0.929	0.460	0.323	0.403
11	Combined M peak & H peak (M ≥ 5 &/or H ≥ 10)	373	166/207	66.15	58.48	0.068a	0.412	0.615	0.525
12	Combined M peak & H peak (M ≥ 5 & H ≥ 10)	294	134/160	76.49	68.28	0.055a	0.391	0.607	0.508
13	Combined M peak & H risen (M ≥ 5 &/or H ≥ 10)	370	166/204	54.58	50.11	0.166	0.403	0.578	0.499
14	Combined M peak & H rise (M ≥ 5 & H ≥ 10)	267	124/143	66.54	62.50	0.239	0.379	0.530	0.460
15	Combined M peak & H peak (M ≥ 10 &/or H ≥ 20)	177	96/81	86.28	85.49	0.674	0.390	0.395	0.392
16	Combined M peak & H rise (M ≥ 10 & H ≥ 20)	149	84/65	77.32	83.17	0.723	0.522	0.346	0.445
17	Combined M peak & H rise (M ≥ 10 &/or H ≥ 20 by 90 min)	279	136/143	42.29	35.71	0.008a	0.382	0.702	0.546

^aP < 0.1 for difference in mean value of lactulose breath test between polyp and non-polyp groups for further assessment.

Rise values are baseline-subtracted peak values during the tests. Bold P values indicate the 7 best models in further assessment. ppm: Parts per million.

Table 5 Model performance with key parameters (area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity)

Panel ID	AUC, % (95%CI)	Accuracy	Sensitivity	Specificity
a	71.6 (66.5-76.7)	0.659	0.666	0.653
b	71.4 (66.3-76.5)	0.642	0.663	0.625
c	72.0 (66.9-77.1)	0.650	0.669	0.634
d	71.7 (66.4-77.0)	0.663	0.679	0.651
e	72.9 (67.3-78.4)	0.651	0.677	0.629
f	72.6 (66.9-78.4)	0.650	0.683	0.622
g	71.7 (65.7-77.7)	0.658	0.673	0.643

AUC: Area under the curve.