Thirty-seven patients with CTS were initially screened. Among them, five patients were excluded because of subclinical ulnar neuropathy noted in the NCS. In total, 32 patients (56 hands) were included in the CTS group. Moreover, 37 healthy volunteers were screened; after asymptomatic hands with abnormal NCS findings were excluded, 32 healthy participants (59 hands) were included in the control group.
Discussion
Our data showed that by using the CSA of the ulnar nerve at the pisiform level as an internal control parameter, MUD and MUR provided higher diagnostic accuracy than SR in the patients with CTS (Table
4). Despite similar AUC, MUD had higher sensitivity (84%) than MUR (63%).
Regarding the application of US for diagnosing CTS, Buchberger et al. first addressed the following four typical ultrasonographic presentations in patients with CTS: A) a flattening of the median nerve at the hamate level; B) an increase in the CSA of the median nerve at the pisiform level; C) an increase in the ratio of median nerve CSA at the pisiform level to median nerve CSA at the distal radius level; and D) palmar bowing of the transverse carpal ligament [
31]. These presentations evolved into various ultrasonographic diagnostic criteria for CTS, such as the FR (the ratio of the length of the long axis of the median nerve to that of the short axis of the median nerve), CSA, and SR methods [
11,
32,
33].
The FR method was later examined in several clinical studies, which have provided controversial results. Using FR > 3.3 as the diagnostic criterion, Duncan et al. compared FR between patients with CTS and normal participants, and they indicated a low likelihood ratio of 1.53 [
34]. Mallouhi et al. used FR > 3 as the cut-off value and obtained sensitivity of 60% and specificity of 76% [
35]. One recent study showed that FR does not significantly differ between symptomatic and asymptomatic groups and shows no correlation to electrodiagnostic testing [
18]. Consistent with these findings, FR in our study did not significantly differ between the healthy volunteers and patients with CTS.
The CSA method initially showed suboptimal accuracy compared with the other methods, possibly because of indirect measurement by assuming that the entire median nerve CSA is elliptical in shape [(long axis × short axis × π)/4] [
31,
36]. Sensitivity of the CSA method improved after a direct measurement method (continuous tracing of the inner border of the endoneurium) was used, and the CSA method was correlated with electrodiagnostic studies [
11‐
13]. Our study revealed that median nerve CSA at the pisiform level significantly differed between the healthy volunteers and patients with CTS, and it provided fair accuracy when 10.35 mm
2 was used as the diagnostic value.
The use of one definitive cut-off value of median nerve CSA to diagnose CTS has been questioned for various reasons. First, CSA may be affected by several demographic factors, such as body height, body weight, age, and sex [
14,
37]. According to one study, healthy volunteers with higher body height, higher body weight, and male sex have larger nerve CSA in the upper limb nerves (median nerve, ulnar nerve, radial nerve, and musculocutaneous nerve) [
38]. Zaidman et al. also showed that with an increase of 1 cm in the body height of a healthy volunteer, median nerve CSA increased by 0.06 mm
2, [
37]. Second, nerve CSA might differ among ethnicities. For example, some studies have demonstrated that the mean median nerve CSA of the American population is larger than that of the Korean population [
14,
38].
Therefore, a comparison of the ratio between two levels of the median nerve, such as in the SR method, was introduced to avoid potential bias caused by the aforementioned individual variations. The SR method compares median nerve CSA between the pisiform level and radioulnar joint level. This method has been reported to have sensitivity of 66.7% for diagnosing clinical symptoms in patients with CTS [
10]. However, for the SR method, our study revealed relatively low sensitivity (53%) and AUC (0.61) when ≥1.32 was used as the cut-off value. The SR method also has limitations for detecting more severe CTS [
39]. This might be explained by the phenomenon of concurrent swelling of the median nerve proximal to the entrapment site. As previously reported, proximal swelling of the median nerve beyond the pisiform level was noted in patients with moderate CTS [
40], which may lead to a reduction in median nerve CSA differences between the pisiform level and the DRUJ level. Consequently, it might undermine the diagnostic accuracy of the SR method. This phenomenon of concurrent proximal swelling of the median nerve was also noted in our study; as shown in Table
3, CSAs of the median nerve were significantly larger in the patients with CTS than in the healthy volunteers, not only at the pisiform level but also at the DRUJ level.
The WFR method compares the median nerve at the pisiform level to the forearm level, ranging from 4 to 12 cm proximal to the distal wrist crease according to different studies [
41]. Although attempts have been made to exclude the potential proximal swelling of the median nerve at the DRUJ level in this method, a significant difference in median nerve CSA at the forearm level between patients with CTS and healthy control has still been noted in previous studies
6,17, thus undermining the role of the forearm median nerve as an internal control [
6,
17]. Furthermore, dense connective tissues are present around the median nerve at the forearm, which occasionally renders identification of the true boundary of the epineurium challenging [
42].
Using ulnar nerve CSA at the pisiform level as an internal control and comparing it with median nerve CSA can avoid the confounding effects of proximal median nerve swelling. Eom et al. showed that there was no change in ulnar nerve CSA between patients with CTS and healthy controls; they also demonstrated a strong correlation between MUR and the electrophysiologic stage of CTS [
19]. Furthermore, Yurdakul et al. demonstrated that compared with the CSA and SR methods, the MUR method was the only method that positively correlated with symptom duration [
21]. Based on these results, the ulnar nerve may be a suitable internal control for evaluating CTS. However, in previous studies, the MUR method did not show superior diagnostic value to the CSA method [
20,
21]. In our study, the MUR method showed slightly higher AUC than the CSA method only (0.75 vs 0.70). Therefore, we introduced the MUD method, which compared the difference between the CSA of the median nerve and the CSA of the ulnar nerve at the pisiform level, as an alternative parameter to assist in diagnosing CTS. Klauser et al. introduced the wrist-to-forearm nerve difference (WFD), which is the difference between the CSAs of the median nerve at the carpal tunnel and forearm level [
41]. It showed higher sensitivity and specificity than the CSA method. Compared with the WFR method, WFD also had a stronger correlation with CTS patients’ clinical severity despite similar AUC [
43]. Our results also revealed the similar AUC of MUD and MUR (0.78 vs 0.75). However, MUD had the highest sensitivity (84%) compared with MUR, median nerve CSA at the pisiform level, and SR (sensitivity: 63, 63, 53%, respectively); thus, MUD is useful in screening CTS using US.
This study has some limitations that should be considered. First, we only recruited patients who met the clinical criteria of idiopathic CTS. This might not reflect the reality of daily clinical practice, where patients may have median neuropathy at the wrist level that may be secondary to various types of underlying conditions. Second, the inclusion criteria of this study included positive results of Tinel’s or Phalen’s test, which have moderate sensitivity. Moreover, we did not use the mixed-nerve short-segment comparative test as a diagnostic criterion in the NCS. Thus, the patients we recruited were more severely affected, and 59% of the patients had moderate to severe CTS (41% of these patients had slow sensory nerve conduction velocity and prolonged distal motor latency, and sensory or motor responses were absent in 18% of the patients). Therefore, further study may be required to explore the sensitivity and specificity of the MUR method in patients with CTS who exhibit mild symptoms. Third, the MUR method was based on the assumption that the ulnar nerve at wrist level is less likely to have compressive neuropathy; hence, it could be used as an internal control. Consequently, this method cannot be applied to patients with clinical symptoms of ulnar neuropathy at the wrist level.