Keywords
anatomic location - carpal tunnel - safety - steroid injection
Introduction
Injection of corticosteroids into the carpal canal is a well-documented therapeutic
and diagnostic procedure in the nonoperative management of mild-to-moderate carpal
tunnel syndrome.[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10] Although the procedure is commonly performed and has been shown to be relatively
safe, complications may still arise. Improper needle placement can result in injury
to the median nerve (MN), ulnar neurovascular bundle (UB), or radial artery (RA),
leading to exacerbation of symptoms, permanent neurologic deficit, or vascular injury.
A variety of recommendations for injection techniques have been made to minimize injury,
including injection medial (ulnar) to the palmaris longus (PL) tendon,[9]
[11]
[12] in line with the fourth digit,[5]
[13]
[14] between the PL and flexor carpi radialis (FCR) tendons,[15]
[16] between the PL and flexor carpi ulnaris (FCU) tendons,[13] and through the FCR tendon.[14] While injections have been shown to be successful in many cases, success is not
uniform. One possible explanation for variability in clinical outcomes is inconsistent
placement of the injection and improper dispersion of fluid within the carpal canal
as a result of utilizing superficial landmarks alone. A better understanding of the
anatomical relationship of the underlying structures based on total wrist width could
result in more consistent delivery of fluid, while minimizing potential complications.
This study was designed to analyze the safety of a volar radial (VR) and volar ulnar
(VU) injection as a function of total wrist width. Although prior studies have utilized
the aforementioned landmarks to determine injection location, there is often high
variability in these landmarks with respect to the underlying structures. A standardized
anatomical technique based on wrist width could further reduce risk to surrounding
structures and be easily applied in the clinical setting with no associated increase
in cost to the patient or utilization of resources to the provider.
Materials and Methods
A total of 87 cadaveric upper extremities with no evidence of previous injury or disease
to the distal forearm or wrist were used to simulate carpal tunnel injections. Thirty-six
wrists were initially dissected to determine the locations of the neurovascular structures
and tendons at the level of the distal wrist flexion crease. Total wrist width, defined
as the distance from the skin overlying the radial styloid extending to the skin overlying
the ulnar styloid, was recorded for each specimen. The overlying skin and portions
of underlying forearm fascia were then removed at this level, exposing the volar wrist
and forearm structures ([Fig. 1]). Dissection of the underlying soft tissues continued until all structures were
fully exposed. Starting at the radial aspect of the wrist, the distances from the
radial styloid to the center of the RA, FCR tendon, PL tendon, MN, UB, and FCU tendon
were measured. Specimen laterality (right vs. left wrists) and gender were recorded.
All measurements were taken using a caliper (General Tool Company, Cincinnati, Ohio,
United States) and measured in millimeters (mm; accuracy, 0.01 mm).
Fig. 1 A view of the volar right wrist following removal of a portion of the skin. A caliper
is used to measure the total wrist width (in mm) and the distance of the tendons and
neurovascular structures from the radial aspect of the wrist. The probe shows the
location of and how measurement of the median nerve was taken.
After documenting the locations of the anatomical structures located at the level
of the carpal canal, 32 additional wrists were used to simulate injections and determine
the location of the needles in relation to the neurovascular structures. Two 18-gauge
hypodermic needles were inserted perpendicular to the wrist at or just proximal to
the distal wrist flexion crease, simulating both a VR- and VU-sided carpal tunnel
injection into each wrist ([Fig. 2]). The first needle was placed into the radial aspect of the wrist at a location
reproducing a VR wrist injection, while the second needle was placed at a location
reproducing a VU injection. It is important to note that the locations of these needles
were based on the previous dissection measurements as a function of total wrist width
to attempt to minimize damage to the underlying neurovascular structures.
Fig. 2 Figure showing placement of two 18-gauge needles, one at 33% and the other at 66%
percent of the wrist width as measured from the radial aspect of the wrist.
To reduce inadvertent penetration of these structures, the VR injections were placed
at a measured location between the RA and MN, which was calculated as 30 to 33% of
total wrist width. The VU injections, on the other hand, were placed at a measured
location between the MN and UB, which again was calculated from our prior findings
and found to be 60 to 66% of total wrist width. All measurements were made from the
radial styloid, using the same calipers and previously described techniques. Wrist
dissections were performed leaving the needles in situ, thus allowing the locations
of the RA, MN, and UB to be recorded relative to needle position. Any penetration
of these structures was noted.
To determine the efficacy of these injection sites, a final 19 fresh-frozen cadavers
were divided into two groups: VR versus VU injection groups. Carpal tunnel injections
were delivered via a 1.5-inch, 22-gauge needle into the canal by one of two techniques
by a single injector (D. H. L.). Each injection contained 1 mL of bacteriostatic normal
saline and 1 mL of Omnipaque 180 radio-opaque dye (GE Healthcare, Princeton, New Jersey,
United States). The VR technique (10 wrists) was performed using 30% of the total
wrist width from the radial styloid. The VU technique (nine wrists) was performed
using 60% of the total wrist width. Both injections were performed at the level of
or slightly proximal to the wrist flexion crease and directed toward the fourth metacarpal
head (radial-sided injection) or index metacarpal head (ulnar-sided injection). Two
milliliters of fluid were injected, and the ease of injection was noted. If there
was difficulty or resistance while injecting fluid, the needle was partially withdrawn,
redirected slightly radially or ulnarly, reintroduced, and fluid injected when free,
unobstructed flow was noted.
Following injection, mini C-arm fluoroscopy (OEC Miniview 6800, GE Healthcare-Americas,
Salt Lake City, Utah, United States) was used to obtain three images (anteroposterior,
lateral, and carpal canal view) of each wrist. These fluoroscopic images were later
reviewed by a musculoskeletal radiologist, two attending hand surgeons, and a hand
surgery fellow to determine if the injection had been successfully placed within the
carpal tunnel. All image readers were blinded as to which injection technique was
employed for each set of images.
Statistical Analysis
Specimen characteristics and location of the RA, MN, and UB are reported as range,
mean, and standard deviation (SD) of total wrist width, as appropriate. A Pearson–Clopper
statistical method was used to calculate 95% confidence intervals (CIs) for penetration
of these structures by either the radial or ulnar needles. A paired t-test was performed to compare right versus left, and a two-sample t-test was applied to compare male versus female specimens.
Results
Data from the initial 36 dissections demonstrate that there was a great deal of anatomic
variation among cadaver specimens. The distance of the RA relative to the radial aspect
of the wrist ranged from 8 to 28% of total wrist width (mean: 17%; SD: 4%; 95% CI:
15.4–18.2%), MN 31 to 59% (mean: 46%; SD: 7%; 95% CI: 43.8–48.2%), and UB 63 to 86%
(mean: 72%; SD: 6%; 95% CI: 69.6–73.6%), respectively. Male specimens demonstrated
a statistically significant larger total wrist width compared with female specimens
(male: 63.8 mm; female: 58.7 mm; p = 0.0004). However, analysis of the locations of the RA, MN, and UB between male
and female specimens as a function of total wrist width showed no statistical significance
(RA: p = 0.91; MN: p = 0.86; UB: p = 0.24). Right-sided specimens were slightly larger than those on the left, but this
difference was also not statistically significant (mean: 0.22 mm; p = 0.783).
The MN averaged 8.12 mm in width at the level of the carpal tunnel (range: 6–14; SD:
1.61; 95% CI: 7.59–8.65). The PL was ulnar to the MN in 19 subjects (N = 53%; range: 1–31 mm; mean distance: 4.21 mm; SD: 6.71 mm; 95% CI: 1.19–7.23 mm),
radial in 9 subjects (N = 25%; range: 1–8 mm; mean distance: 5.0 mm; SD: 2.06 mm; 95% CI: 3.65–6.35 mm),
directly overlying the MN in 4 subjects (N = 11%), and not present in 4 subjects (N = 11%). The FCR was located radial to the MN in all patients and the distance ranged
from 2 to 17 mm (mean: 9.36 mm; SD: 3.63 mm; 95% CI: 8.18–10.55 mm). There was also
a wide range of variation in the locations of the superficial landmarks relative to
total wrist width: FCR 19 to 46%; PL 31 to 95%; and FCU 71 to 93%.
From these initial results, it was determined radial-sided injections would be placed
at 30%, while ulnar-sided injections would be placed at 60%, as both were found to
be “safe zones” in our previous dissections. We additionally evaluated one more radial
(33%) and ulnar (66%) location to determine the effects of moving the injections closer
to the MN and UB. Using a radial-sided approach at 30 or 33%, there were no penetrations
of either the RA or MN. On the other hand, an ulnar approach at 60% demonstrated two
penetrations of the MN and six penetrations of the UB at 66%. The ulnar artery was
penetrated in four specimens and ulnar nerve in one specimen, and the needle was located
between the ulnar nerve and artery in one specimen.
In the final 19 specimens, all injections (19/19) were successfully placed in the
carpal canal. A total of 9 were injected from the ulnar side, while 10 were injected
from a radial approach. Accurate placement of all injections was confirmed in three-plane
fluoroscopy. Each ulnar-sided injection resulted in unobstructed flow with little
resistance and showed wide dye dispersion on imaging ([Fig. 3A–C]). In four of the radial-sided injections through or near the FCR tendon, there was
initially some resistance to the injection requiring redirection of the needle until
unobstructed flow was obtained. Fluoroscopic imaging showed dye primarily deposited
on the radial side of the canal, and in 8 of the 10 specimens (80%), a detectable,
often substantial amount of dye was found tracking down the flexor pollicis longus
(FPL) tendon sheath into the thumb ([Fig. 4A–D]).
Fig. 3 Fluoroscopic (A) anteroposterior, (B) lateral, and (C) carpal tunnel images after an ulnar-sided injection of the carpal canal showing
radio-opaque dye located within the carpal canal.
Fig. 4 Fluoroscopic (A) anteroposterior, (B) lateral, and (C) carpal tunnel images after a radial-sided injection of the carpal canal showing
radio-opaque dye located within the carpal canal. (D) Radio-opaque dye shown along the flexor pollicis longus flexor tendon sheath.
Discussion
Carpal tunnel injection is a valuable procedure for diagnostic and therapeutic use
in the treatment of carpal tunnel syndrome and is commonly utilized by a wide range
of health care providers. Our cadaveric analysis of 87 specimens comparing safety
between a VR- and VU-sided injections as a function of total wrist width demonstrates
that a VR approach results in a slightly lower incidence of iatrogenic injury to the
underlying neurovascular structures compared with a VU approach. There was a wide
range of anatomic variation in the relationship between the RA, MN, UB, and surrounding
soft tissue structures among specimens, indicating that the use of superficial landmarks
alone may result in an increased risk of iatrogenic injury and variability in clinical
outcomes. Our standardized anatomical approach based on total wrist width demonstrated
significantly less variability than prior techniques.
The risk of iatrogenic injection into the underlying neurovascular structures resulting
in exacerbation of symptoms, permanent neurologic deficit, or vascular injury is a
known potential complication of this procedure. Various authors have performed studies
and recommended techniques based on safety. Gelberman et al[15] determined the MN to be 6-mm wide and 2.1-mm thick at a location 1 cm proximal to
the distal wrist flexion crease and advocated for injection through the FCR tendon,
with further studies by Dubert and Racasan[17] and Racasan and Dubert[14] supporting this approach. Moving ulnarly, Graham et al[4] suggested needle placement just ulnar to the FCR, and Koo and Szabo[18] advised injection between the FCR and PL. Additional recommended techniques include
needle placement ulnar to the PL,[1]
[5]
[8]
[9]
[12]
[19] as well as between the PL and FCU.[13]
[15]
[20]
We argue, however, that due to differences in patient size, variations in location
of neurovascular structures relative to superficial landmarks, and the wide range
of superficial landmarks relative to total wrist width observed in the present study,
a standardized technique based on total wrist width may result in a lower incidence
of iatrogenic injury. Our results demonstrate that superficial landmarks vary widely
among patients. For example, the PL was located ulnar to the MN in 53% of patients
with a mean distance of 4.21 mm (range, 1–31 mm), radial in 25% with a mean distance
of 5.0 mm (range, 1–8 mm), directly overlying the MN in 11%, and not present in 11%
of subjects. These findings indicate that a carpal tunnel injection placed just ulnar
to the PL as described above could potentially result in inadvertent damage to the
MN in 36% of patients who have a PL either directly overlying or radial to the MN.
Our data also suggest that any technique that relies on the PL as a superficial landmark
cannot be used in 11% of patients who are without this muscle.
Although such wide variation among superficial landmarks exists, our results regarding
the locations of the underlying neurovascular structures as a function of total wrist
width are consistent regardless of patient gender or wrist laterality. Analysis of
the locations of the RA, MN, and UB between male and female specimens showed no statistically
significant differences (RA: p = 0.91; MN: p = 0.86; UB: p = 0.24) based on percentage of wrist width. Our findings are further generalizable
in that there was minimal difference in wrist width between right and left specimens.
Although right wrists were, on average, slightly larger than those on the left, this
difference was not significant (p = 0.783). By measuring total wrist width and using a percentage of this value to
determine the injection site, we have attempted to provide guidelines for carpal tunnel
injections that can be easily reproduced regardless of variation in a patient's superficial
anatomy.
Moreover, the reproducibility of injection delivery has been infrequently evaluated
and results have varied. Minamikawa et al[21] performed a series of carpal tunnel injections in 16 fresh cadaver forearms using
methylene blue to compare the efficacy of fluid delivery while also minimizing risk
of intraneural injection. The specimens were dissected and dye diffusion was quantified
among four groups. They demonstrated the best diffusion was achieved when 2 mL of
methylene dye was injected 3 cm proximal to the distal wrist flexion crease ulnar
to the PL in line with the third web space. This technique was found to be superior
to more radial-sided injections located between the PL and FCR, or injections performed
1 cm proximal to the distal wrist flexion crease. In one radial-sided approach (6%),
the carpal tunnel was missed entirely. Wood[12] demonstrated that an experienced injector missed the carpal tunnel 8% of the time
(2/26 wrists). However, this study involved placing needles in the standard injection
position and dissecting the wrist to find the tip position. No fluid was injected,
and their study rationalized that the probable miss rate would be lower in clinical
practice as the surgeon would detect resistance to injection and reposition the needle
accordingly.
Our results suggest the carpal tunnel may be reliably injected from either the VR
or VU approach. Radial-sided injections placed at either 30 or 33% of total wrist
width resulted in no penetrations of the RA or MN. With the VR approach, one can successfully
inject fluid into the carpal tunnel with a low risk of iatrogenic injury to the underlying
neurovascular structures, but the injector may expect to encounter some soft tissue
resistance to the injection, as well as the unanticipated effect of injecting fluid
into the FPL tendon sheath. Inadvertent injection into the FPL tendon sheath occurred
80% of the time with a radial-sided approach. Alternatively, penetration of critical
structures using VU injections resulted in two penetrations of the MN at 60% of total
wrist width and six penetrations into the UB at 66%. The VU injection showed equal
efficacy in fluid delivery within the carpal canal, had less soft tissue resistance
to the injection, and eliminated the likelihood of injecting into the FPL tendon sheath.
While this study resulted in novel findings and proposed a new standardized approach
for carpal tunnel injections based on total wrist width, the experimental design had
its limitations. The study was based on a cadaveric model, and as such, lacked clinical
outcomes' data produced by similar retrospective or prospective trials. This pilot
study, however, was necessary to provide preliminary anatomical data, which will be
used to design a more rigorous prospective study. Another inherent limitation based
on a cadaveric model is that intraneural injection cannot be easily detected in the
absence of patient feedback. In the clinical setting, an intraneural injection into
a median or ulnar nerve would likely produce pain and/or paresthesias, acting as an
immediate indication of improper needle placement. The location and direction of the
needle would then be appropriately adjusted, potentially altering outcomes. However,
a patient may misinterpret these symptoms as simply being part of the injection itself,
unable to differentiate pain from the injection versus the needle being placed intraneurally.
Also, in certain situations (e.g., a sedated patient undergoing a carpal tunnel injection
for a carpal tunnel release), this sensory feedback may not be present. Using a standardized
approach would be helpful as a starting point for carpal tunnel injection regardless
of patient size or variations in surface anatomy.
Although carpal tunnel injections are a relatively safe and effective procedure, the
ideal location to minimize risk of iatrogenic injury to the underlying RA, MN, and
UB remains controversial in the literature. Our data demonstrate a wide variability
in commonly described superficial landmarks, such as the FCR, PL, and FCU. This variability
may be responsible for inconsistent placement of the injection within the carpal canal
and an increased risk of adverse events. While further prospective studies are warranted
to more rigorously evaluate the efficacy and clinical outcomes of these guidelines,
our data demonstrate radial-sided injections placed at one-third of total wrist width,
as measured from the radial styloid, offer a safe and reliable standardized approach
to the carpal canal that eliminates the wide variability associated with superficial
musculature. Our results also demonstrate an ulnar-sided injection using 60% of total
wrist width is relatively safe, although moving further in an ulnar direction should
be avoided to prevent injury to the UB.
