3.1 Hp levels in lung cancer and CAP patients

A potential difference in serum Hp levels between lung cancer and CAP patients was examined. In detail, serum Hp levels were measured in 111 lung cancer patients distributed over a wide range of ages (age distribution: 41 to 85 years) and 80 CAP patients (age distribution: 41 to 85 years) and subjected to receiver operating characteristic (ROC) curve plotting. Concentration measurements are graphically shown in Figure 1 and statistically summarised in Table 3. Area under the ROC curve (AUC) values were calculated by the partial integration of the ROC curve; sensitivity and specificity values are summarised in Table 4.

	NM	LCM	CAPM	NF	LCF	CAPF	NMF	LCMF	CAPMF
Number of values	50	50	40	50	61	40	100	111	78
25% percentile	0.488	0.928	0.939	0.585	0.965	1.18	0.546	0.958	1.08
Median	0.676	1.20	1.22	0.702	1.20	1.61	0.682	1.20	1.46
75% percentile	0.904	1.42	1.80	0.984	1.58	1.93	0.914	1.49	1.90
Mean	0.693	1.23	1.40	0.774	1.29	1.61	0.733	1.26	1.51
Std. deviation	0.267	0.421	0.575	0.337	0.502	0.482	0.305	0.467	0.537
Std. error	0.038	0.060	0.092	0.048	0.064	0.077	0.031	0.045	0.061

NM, normal male control; LCM, lung cancer male; CAPM, CAP male; NF, normal female control; LCF, lung cancer female; CAPF, CAP female; NMF, normal male/female control; LCMF, lung cancer male/female; CAPMF, CAP male/female

	Lung cancer			CAP
	LCM	LCF	LCMF	CAPM	CAPF	CAPMF
AUC ± SEa	0.878 ± 0.033	0.820 ± 0.040	0.860 ± 0.028	0.865 ± 0.040	0.905 ± 0.037	0.882 ± 0.027
Sensitivity (%)	75.5	78.7	76.6	72.5	82.5	81.2
Specificity (%)	84.0	73.5	79.8	88.0	89.8	83.8
Cut-off value(A450nm)	0.938	0.954	0.954	0.958	1.136	0.995

Hp measurements of the normal male and female controls used as the reference

LCM, lung cancer male; LCF, lung cancer female; LCMF, lung cancer male/female; CAPM, CAP male; CAPF, CAP female; CAPMF, CAP male/female; AUC, area under the ROC curve; SE, standard error of the mean

^ap < 0.0001

As seen in Figure 1 and Table 3, the mean Hp concentrations in the normal male/female control (NMF) group, the lung cancer male/female (LCMF) group, and the CAP male/female (CAPMF) group were 0.733, 1.26, and 1.51, respectively. This shows that the serum Hp levels increased by approximately 72% and 106% in LCMF and CAPMF patients, respectively, compared to NMF subjects, indicating that the Hp level is approximately 20% higher in CAP patients than in lung cancer patients. The difference in the elevation of Hp levels in the lung cancer versus CAP groups was more profound in the female groups (LCF versus CAPF:∼25%) than that in the male groups (LCM versus CAPM:∼14%). As indicated in Figure 1, the difference in the serum Hp level between male patients with lung cancer and CAP is not statistically significant (p > 0.05), whereas the difference between female patients is evident (p = 0.002).

For the LCM and CAPM groups compared to the NM subjects, the AUC values of Hp were measured to be, respectively, 0.878 and 0.865, with a sensitivity of 75.5% and 72.5% and a specificity of 84.0% and 88.0% (Table 4). For the LCF and CAPF groups compared to the NF subjects, AUC values of Hp were, respectively, 0.820 and 0.905, with a sensitivity of 78.7% and 82.5% and a specificity of 73.5% and 89.8%.

These data show that Hp can be used to screen for lung cancer and CAP, as displayed in Table 4 (AUC values for LCMF: 0.860, those for CAPMF: 0.882), although Hp has a disadvantage in its selectivity due to simultaneous elevations of Hp levels in both patient groups. It is worth noting that the selectivity between female lung cancer and CAP patients is superior compared to male patients.

3.2 Trx1 levels in patients with CAP and various diseases

We then examined the difference in serum Trx1 levels between patients with CAP and those with lung cancer. We aimed to investigate serum Trx1 levels due to reports that the intracellular synthesis of Trx1 increases in response to cancerous processes [5]. Serum Trx1 levels may also be increased in response to lung carcinoma progression. We selected lung cancer as a representative cancer for this study because CAP is an inflammatory disease of the lung. Serum Trx1 levels were measured by ELISA in the patients, as well as in normal male and female controls.

As seen in Figure 2, the mean concentrations of Trx1 in the NMF, CAPMF, and LCMF groups were 0.504, 0.417, and 0.568, respectively, showing that the serum levels of Trx1 decreased by approximately 21% in the CAPMF patients. Trx1 levels increased by approximately 13% in the LCMF group compared to the NMF group. In addition, Figure 2B shows that serum Trx1 levels increased along with lung cancer progression, indicating that the increase of Trx1 levels may be caused by the cancerous process. The data for NF, NM, LCM, and LCF subjects are not separately shown here because of their similarity with the data for the NMF and LCMF groups.

Interestingly, serum Trx1 levels were at remarkably lower values in CAPF patients (0.360) than in CAPM patients (0.457), although the Trx1 levels are nearly the same in the respective controls. This finding indicated that the serum levels of Trx1 decrease by approximately 27% and 8% in CAPF and CAPM patients, respectively, compared to their corresponding normal controls (NF: 0.490; NM: 0.495) (Figure 2A).

Serum Trx1 levels were also measured in female patients with rheumatoid arthritis (Arth F) and cardiovascular disease (Cardio F). We chose to look at these groups because the intracellular synthesis of Trx1 is increased in response to oxidative stress, so we reasoned that serum Trx1 levels may also be increased in response to inflammatory and oxidative stress-induced diseases. As seen in Figure 2A, the mean serum Trx1 concentrations in the Arth F and Cardio F groups were 0.485 and 0.552, respectively, indicating that Trx1 levels in the Arth F patients are nearly the same as the levels in the NF group (p = 0.761). In contrast, the Trx1 levels increased by approximately 3% in Cardio F patients compared to the NF group (p = 0.001). The opposite trends observed for the serum Trx1 levels between CAP and the other investigated diseases support the utility of Trx1 as a diagnostic marker for CAP over Hp because serum Hp levels are significantly elevated in both patients with CAP and those with lung cancer.

Taken together, the data described above indicate that Trx1 may be useful as a promising CAP marker, especially for female subjects, because it is capable of clearly identifying CAP patients, whether they have been affected by cancer or another disease.

In addition, ROC curve analysis was performed for the Trx1 levels of the cardiovascular group, with the increased Trx1 values compared to normal female values. For the Cardio F group compared to the NF group, the AUC value of Trx1 was measured to be 0.677 ± 0.063, indicating that this type of oxidative stress-related disease apparently increases Trx1 levels.

Taken together, the results described above indicate that Trx1 can discriminate between the female CAP group and the normal female controls at a probability of about as high as 88% and can be used as a CAP marker with superior sensitivity and specificity.

3.3 Reverse correlation of Trx1 levels with Hp levels

As is understood from the data, Hp is a good marker for both male and female patients with CAP, whereas Trx1 exhibits a function similar to Hp in female patients. The characteristic difference between Hp and Trx1 is that the CAP group has a higher serum Hp level but a lower serum Trx1 level compared to the corresponding normal controls. To confirm these contrary results, the correlations between the Trx1 measurements and the Hp measurements of the CAP group were analysed, and the results are shown in Figure 3. The Trx1 measurements were found to be inversely proportional to the Hp measurements, (r = −0.231, p = 0.041), as seen in Figure 3. For testing for normality of the data, the D’Agostino-Pearson normality test was performed. The data sets for Trx1 and Hp levels were normally distributed (Trx1: p = 0.06; Hp: p = 0.30).