Introduction
The most effective and durable anti-reflux intervention is a laparoscopic Nissen fundoplication
(LNF) [1]
[2]
[3]. However, alternative treatments for gastroesophageal reflux disease (GERD) have
emerged over the last 15 years, including transoral incisionless fundoplication (TIF).
It is a safe and effective minimally invasive endoscopic technique for the management
of GERD in selected patients with small hiatal hernias ≤ 2 cm and small diaphragmatic
defects (Hill grade 1–2) [4]
[5]
[6]
[7]. Introduced in 2005, TIF is an endoscopic procedure that creates a flap valve with
full-thickness serosa-serosa plications of the esophagus and gastric cardia, with
the aim of restoring the angle of His, similar to a surgical partial fundoplication
[6]
[8]. Multiple randomized controlled clinical trials report resolution of troublesome
regurgitation not responding to proton pump inhibitor (PPI) therapy in the majority
of patients treated with TIF [4]
[5]
[9]
[10]
[11]
[12]. In particular, a randomized controlled trial published in 2015 by Trad et al (the
TEMPO trial) showed that TIF was more effective than PPI therapy for refractory GERD
symptoms [5]. The follow-up study reported durable symptom resolution of 88 % at 1 year, 90 %
at 3 years, and 88 % at 5 years [13]. Furthermore, a systematic review and meta-analysis performed by McCarty et al of
32 studies (involving 1475 patients) concluded that TIF is a technically feasible
and effective endoscopic procedure for the management of GERD with a technical success
rate of 99 % and low rate (2 %) of serious adverse effects [11]. Another systematic review with network meta-analysis showed that TIF is associated
with a higher likelihood of improved health-related quality of life compared to PPI
and laparoscopic Nissen fundoplication [14].
In the published clinical trials, experienced surgeons and gastroenterologists performed
TIF [4]
[5]
[12]
[13]
[15]
[16]. There are very limited data on the learning curve of TIF when performed by a gastroenterologist.
We present a single gastroenterologist’s early experience to study the learning curve
of TIF and propose threshold procedure volumes to achieve proficiency and efficiency.
Patients and methods
This was an observational study conducted in a single American tertiary care academic
center. Data were collected prospectively on consecutive patients referred to the
Johns Hopkins Heartburn Center who had a TIF procedure performed by an experienced
therapeutic endoscopist (MC) between September 2017 and February 2020. All patients
provided written informed consent to participate in the Johns Hopkins Heartburn Center
(TIF) Registry, which prospectively tracks patients undergoing TIF at the Johns Hopkins
Hospital (Johns Hopkins IRB number IRB00201157; clinicaltrials.gov NCT03853772). Prior
to performing TIF without supervision, the gastroenterologist acquired cognitive knowledge
and technical training on performance of TIF by review of a standardized curriculum
in a classroom-type setting, hands-on performance of TIF on an inanimate and live
animal model, and in-person observation of cases performed by a highly experienced
gastroenterologist who achieved expertise in TIF (KC). These observed were not included
in the TIF learning curve.
Currently, TIF using the 2.0 technique is performed with the EsophyX-Z or Esophyx-Z + device
(EndoGastric Solutions, Redmond, WA) under general anesthesia with naso- or endotracheal
intubation ([Fig. 1]) [17]. The EsophyX device with videoendoscope inserted is advanced into the stomach, and
endoscopic fundoplication is achieved in the retroflex position. Small hiatal hernias
can be reduced by using a helical retractor that is inserted into the gastroesophageal
junction and gently pulled down the esophagus below the diaphragm while decompressing
the stomach. The fundoplication is created through clockwise and counterclockwise
wrapping of tissue at the gastric cardia at the 11 o’clock and 1 o’clock positions
and placement of multiple transmural esophago-gastric plications using polypropylene
H-fasteners with comparable strength to 3.0 sutures placed to create a fundoplication
that is at least 270 degrees in circumference [17]. Valve length is achieved by withdrawing the esophagus from 2 to 4 cm into the stomach
and placing fasteners at intervals between the 11 and 1 o’clock positions.
Fig. 1 Endoscopic images of transoral incisionless fundoplication using TIF 2.0 technique.
a Endoscopic image of the Esophyx-Z + device seen in retroflex view during the initial
step of transoral incisionless fundoplication. The helix (arrow) is inserted into
the gastric cardia side of the squamocolumnar junction, and with concomitant suctioning,
the distal esophagus is withdrawn in caudad direction under the diaphragm to create
a flap valve 2 to 3 cm in length. The distal esophagus and gastric cardia are shown
enclosed within the tissue mold prior to placement of plastic H fasteners to create
a full thickness esophago-gastric plication. In TIF 2.0 technique, at least 20 fasteners
are placed at different locations to create a flap valve that is at least 270 degrees
in length. b The endoscopic image on the right is of the TIF valve 6 months later and shows the
gastric cardia snug around the endoscope (normal appearance, Hill grade 1), with a
310-degree circumference.
Routine pre-TIF work-up included upper endoscopy (EGD) with biopsy and careful evaluation
of esophageal landmarks, cine-esophagogram, high-resolution esophageal manometry (HREM)
(in patients with dysphagia) and ambulatory pH-monitoring with either 4-day Bravo
wireless pH capsule off medications or 24-hour pH impedance. Patients were considered
appropriate for TIF if they had an abnormal pH-test (acid exposure time > 6 %), hiatal
hernia ≤ 2 cm, Hill grade ≤ 2, and a BMI < 35 [5]. Patients who had per-oral endoscopic myotomy (POEM) prior to TIF were excluded
from the analysis. TIF was performed using the EsophyXZ or EsophyX Z + device and
the TIF 2.0 protocol [17].
Patient demographic information, axial length of any hiatal hernia, Hill classification
of the gastroesophageal flap valve [18], TIF procedure indication, location and number of plications, total number of fasteners
placed, TIF valve circumferential extent, TIF valve length, and TIF procedure time
(time from first passage to withdrawal of the EsophyX device) were recorded on a secure
online research electronic data capture system (REDCap software, Vanderbilt University,
Nashville, Tennessee, United States). Two independent observers who were not part
of the physician or nursing team assessed the final TIF valve length and circumference.
Statistical analysis
The primary endpoint of the study was the threshold number of procedures needed to
achieve proficiency (defined as creation of a TIF valve ≥ 270 degrees in circumference
and > 2 cm in length) in performing TIF. The secondary endpoints were the threshold
number of TIF procedures to achieve efficiency (defined by the minimum mean time to
perform a TIF) and the minimum mean time to perform a plication (place a set of two
fasteners).
Patients and disease characteristics were described as frequencies with percentages
(%), medians [IQR], and means±standard deviation (SD). We used the Cumulative Sum
of Means (CUSUM) analysis to estimate the threshold for achieving proficiency by detecting
sequential changes in achievement rates over time, as the endoscopist gained experience
from each procedure. The essential function of CUSUM is to calculate the trends in
“average success rates” over time. We calculated the average success rate at the completion
of each procedure, and plotted the changes in this rate. In order to detect changes
occuring in various procedure parameters and metrics as the operators experience with
the procedure increased, we checked for structural breaks in the time series data,
i. e. we looked for breakpoints around which the means varied greatly before and after.
Specifically, we determined the number of procedures needed to minimize total TIF
procedure time, time to perform a plication (in minutes), and to consistently achieve
a ≥ 270 circumferential degree of wrap. We performed data analyses and generated figures
using Stata 15 (StataCorp, College Station, Texas, United States). All authors had
access to the study data and reviewed and approved the final manuscript.
Results
A total of 69 patients (63.8 % were males with a mean age of 54 ± 13 years) had 72
TIFs after work-up confirmed appropriate criteria (abnormal pH-metry, < 2 cm hiatal
hernia < 2 cm, Hill grade ≤ 2, BMI < 35). Six patients (8.3 %) had a history of prior
surgical fundoplication. The primary indications for TIF were refractory GERD (44.7 %),
PPI-averse patients (23.6 %), post-ablation of Barrett’s esophagus (11.1 %), laryngopharyngeal
reflux disease (LPRD) (11.1 %), and failed Nissen fundoplication (8.3 %). The median
hiatal hernia axial length was 1 [IQR 0–1.37]; the majority of patients had no hiatal
hernia (37.5 %) or a hiatal hernia measuring 1 cm (33.3 %). The median Hill grade
was 1 [IQR 1–2] ([Table 1]).
Table 1
Patient characteristics.
|
Patients
|
N = 69
|
|
Mean age, SD
|
54 (SD 13)
|
|
Gender, male
|
44(63.8 %)
|
|
Mean BMI, SD
|
27.3 (SD 6.3)
|
|
History of Barrett’s esophagus
|
16 (23.2 %)
|
|
Esophagitis pre-TIF
|
17 (24.6 %)
|
|
LA Grade A
|
8 (11.6 %)
|
|
LA Grade B
|
5 (7.2 %)
|
|
LA Grade C
|
4 (5.8 %)
|
|
Abnormal average acid exposure time (AET)[1]
|
31 (44.9 %)
|
|
Average acid exposure time (AET), median (IQR)
|
6.5 (3.7–8.7)
|
|
Primary Indication for TIF
|
|
|
Refractory GERD
|
30 (41.7 %)
|
|
LPRD
|
8 (11.1 %)
|
|
Failed Nissen
|
6 (8.3 %)
|
|
Failed TIF
|
3 (4.2 %)
|
|
PPI-averse
|
17 (23.6 %)
|
|
Post-ablation BE
|
8 (11.1 %)
|
|
Procedures
|
(n = 72)
|
|
Median hiatal hernia axial length (cm) (at the TIF procedure)
|
1 (IQR 0–1.37)
|
|
Hiatal hernia = 0
|
28 (38.9 %)
|
|
Hiatal hernia = 1 cm
|
25 (34.7 %)
|
|
Hiatal hernia between 1 to 2 cm
|
2 (2.7 %)
|
|
Hiatal hernia = 2 cm
|
17 (23.6 %)
|
|
Median Hill grade (at the TIF procedure)
|
1 (IQR 1–2)
|
|
Hill grade 1
|
42 (58.3 %)
|
|
Hill grade 2
|
27 (37.5 %)
|
|
Hill grade 3
|
3 (4.1 %)
|
SD, standard deviation; BMI, body mass index; IQR, interquartile range; AET, abnormal
average acid exposure time; TIF, transoral incisionless fundoplication; GERD, gastroesophageal
reflux disease; LPRD, laryngopharyngeal reflux disease; PPI, proton pump inhibitor;
BE, Barrett’s esophagus.
1 AET > 4 % off acid suppressive medication.
The overall procedure completion rate was 98.5 %. One TIF procedure (TIF #7) was not
completed due to inability to reduce the hiatal hernia. Subsequent laparoscopic anti-reflux
surgery showed a 2–3 cm hiatal hernia and large crural defect with fat pad, likely
causing underestimation of the hiatal hernia length and width.
All but one patient (admitted for 23 hours for a migraine headache, unrelated to the
TIF procedure) were discharged home. There were no procedure-related serious adverse
events. The mean TIF valve circumference was 290 ± 22.3 degrees. The mean TIF procedure
time was 48.1 ± 13.9 minutes.
Outcomes
Overall, 89 % of all procedures led to the creation of a TIF valve with a minimum
circumference of 270 degrees and minimum length > 2 cm. Notably, three TIFs completed
and performed (TIF # 3,8,15) during the learning phase did not meet the technical
success criteria. In retrospect, these patients met criteria for TIF at a prior EGD
but at the TIF procedure had larger hiatal hernia and/or Hill grade when prior images
were reviewed ([Table 1]). In addition, three patients who had their TIF in the early part of the learning
curve successfully underwent repeat TIF later on due to symptom recurrence. CUSUM
analysis of achievement of technical success revealed an inverted V pattern, indicating
acquired proficiency from the 20th procedure (P = 0.005) ([Fig. 2]). Review of the primary data showed that all TIF procedures were successful and
met target TIF valve parameters from the 18th TIF procedure.
Fig. 2 Cumulative sum (CUSUM) analysis to assess procedure proficiency. (consistent creation
of a ≥ 270-degree and > 2-cm-long TIF valve).
Breakpoint analysis revealed that the threshold to achieve proficiency was achieved
at the 20th procedure (P < 0.0001) ([Fig. 3]), consistent with the CUSUM analysis. Review of the primary procedure data showed
that all TIF valves exceeded the minimum threshold circumference of 270 degrees from
the 18th procedure. With regards to achievement of efficiency, the threshold to improve the
time for each plication was achieved at the 26th procedure (P < 0.0001) ([Fig. 4]). The overall mean time to perform a plication was 3.65 ± 1.56 minutes, which significantly
decreased after achieving proficiency (mean time = 2.8 min ± 0.7 per plication (95 %
CI 2.6–3 minutes), compared to 5.1 ± 1.8 (95 % CI 4.39–5.81 minutes) before 26th procedure, (P < 0.0001). Finally, the threshold to improve the TIF procedure time was noted at
the 44th procedure ([Fig. 5]). The average TIF procedure time gradually decreased with increasing experience,
but after the 44th procedure the mean TIF time decreased significantly to 39.4 minutes (95 % CI 9.4),
compared to 53.7 minutes (95 % CI 13.5) before the breakpoint (P < 0.0001).
Fig. 3 Breakpoint analysis of TIF valve circumference. (Proficiency: consistent creation
of a minimum 270-degree TIF wrap).
Fig. 4 Breakpoint analysis of average plication time.
Fig. 5 Breakpoint analysis of mean TIF time.
Discussion
In this single-center study, we report the early experience of a therapeutic endoscopist
and described for the first time the learning curve for TIF performed entirely in
the endoscopy unit on an outpatient basis. We evaluated several metrics for estimating
thresholds for achieving proficiency and efficiency. All of the metrics revealed two
distinct phases in the learning curve: an early phase and a later phase. The breakpoints
for achieving proficiency and efficiency were different. Proficiency was achieved
after 18 to 20 TIF procedures, efficiency for performing plications was achieved after
26 procedures, and maximum efficiency (reduced TIF time) was achieved after 44 procedures,
suggesting differences in achieving proficiency in various aspects of the procedure.
To our knowledge, this is the first report of the TIF learning curve. In a preliminary
report, Raza et al suggested that TIF could be safely performed by trained therapeutic
endoscopists in the endoscopy suite with no intra-procedure or post-procedure adverse
events [17]. Also, it showed that all patients could safely be discharged within 24 hours post-procedure
[19]. Similarly, our study suggests that TIF can be performed safely and successfully
by an experienced gastroenterologist in the endoscopy unit after a standardized didactic
and observational experience.
Other investigators have investigated the learning curves of complex endoscopic procedures
using different approaches, often including decrease in procedure time. For example,
a single-center retrospective study performed by El Zein et al described the learning
curve for POEM by identifying a threshold number of procedures required to significantly
lower procedure time. With a single experienced therapeutic endoscopist performing
all the procedures, the investigators concluded that a threshold number of 13 POEM
procedures will lead to a learning plateau [20]. Similarly, the learning curve of a single therapeutic endoscopist for endoscopic
sleeve gastroplasty (ESG) using procedure time and number of plications per procedure
showed that these parameters plateau at seven and nine cases, respectively [21]. Both single operator studies suggest a very short, steep learning curve for complex
endoscopic procedures if time is the main parameter to assess learning. In contrast,
the learning curve for therapeutic endoscopic procedures can also be assessed using
defined proficiency benchmarks as well as procedure time. For example, in a large
retrospective study of 1346 patients, Liu et al described the learning curve of POEM
using both procedure time and a composite outcome of technical failure, adverse events,
and clinical failure, suggesting 100 cases were needed, despite prior experience in
endoscopic submucosal dissection (ESD) and a high POEM volume [22].
Other learning curve studies also emphasized achievement of specific goals and not
procedure time. Using a porcine ex-vivo model (Erlangen Active Simulator for Interventional
Endoscopy), Gonzalez et al conducted a prospective ex-vivo learning curve study based
on the number of EUS-FNA procedures performed on a cyst and two solid masses for assessing
learning, such as number of successful attempts, needle view loss, and scope handling
[23]
[24]. Zhang et al performed a retrospective analysis of consecutive ESD to describe the
learning curve of a single operator between 2009 and 2017 [25]. The authors proposed that to achieve proficiency – defined as ≥ 90 % en block resection
and > 80 % margin-negative resection with resection speed of > 9 cm2 per hour – 250 cases are needed. Similarly, ERCP procedure threshold for trainees
has been proposed, based on success of common bile duct cannulation [26].
In the current study, we used a combination of procedure time and technical success
to assess the learning curve for TIF. Using a predefined proficiency benchmark (creation
of a TIF 2.0 valve ≥ 270 degrees and > 2 cm long), we demonstrated a steep rise in
the learning curve, with proficiency achieved after 18 to 20 procedures. When we tracked
proficiency and efficiency using time as the endpoint, we found that with increased
experience, efficiency improves after achieving proficiency and the TIF learning curve
using time plateaus. It is possible that additional experience would further decrease
procedure time and improve efficiency beyond the 72 procedures included in this early
experience.
It is important to mention that TIF is not routinely performed at most medical centers.
It is also not routinely taught in gastroenterology and surgical training programs.
There is a standard set of required reading and ex vivo inanimate model technical
training offered to new users. But beyond this, observation or participation in clinical
cases and supervised learning may vary across different practice settings. It is important
to standardize the learning experience and address potential pathways for training
to enhance clinical adoption of this relatively newer anti-reflux procedure that gastroenterologists
can offer patients with suitable anatomy.
Finally, our early TIF experience underscores the importance of estimating the length
and width of the hiatal hernia. The Hill grade (a better estimate of the hiatal hernia
width and crural defect) [27] is critical factor in the success of TIF without hiatal hernia repair. TIF can typically
reduce small hiatal hernias ≤ 2 cm in length, but larger and wider hernias require
surgical repair and cruroplasty. The hiatal hernia length and width can be underestimated
during endoscopy, as shown in our 4 TIF technical failures. Better methods for accurate
estimation of the hiatal hernia size, standardization of the best method and reporting
of findings, and widespread education of endoscopists on the proper assessment of
the anti-reflux barrier are needed.
The strengths of our study include the documentation of the learning experience prior
to independent procedure performance, prospective data collection, use of time as
well as objective procedure metrics, independent observer assessment of the endpoints
(TIF valve characteristics), and quantitative analyses using two statistical approaches
demonstrating consistency in our results. However, there are also certain limitations
to our study. First, it was a single-center study and reports a single endoscopist’s
experience. Hence, generalizability is limited. There is certainly going to be some
variability in the learning curves and number of TIF procedures needed to achieve
technical success among different operators. It would be interesting to study the
learning curve of different specialties and experience levels, such as trainees, general
surgeons, specialized foregut surgeons, and therapeutic endoscopists. Second, our
study included TIF data on a limited number of patients, but our sample size of 72
TIF procedures approaches or exceeds that in some multicenter randomized controlled
trials where TIF was performed by more than one operator [5]
[6]
[12]
[28]. Finally, we report the learning curve for TIF only considering proficiency and
efficiency. Larger multicenter studies including hundreds of procedures that assess
the learning of TIF considering all aspects (technical success, safety, and clinical
outcomes), similar to the study reported for the learning curve of radiofrequency
ablation [29], are needed. These studies are already underway (NCT03853772). Moreover, it would
be important to understand prior operator experience and assess its impact on safety
and efficacy in future clinical trials involving TIF, because early clinical trials
did not routinely document this. Nevertheless, despite these study limitations, this
is the first prospective study of the learning curve for TIF 2.0, which provides a
basic understanding of procedural experience needed to potentially assist future research
studies, training programs, and hospital credentialing committees.
Conclusions
In conclusion, TIF can be safely, successfully, and efficiently performed in the endoscopy
suite by a gastroenterologist. Proficiency can be achieved after 18 to 20 procedures,
while efficiency is optimal after the 44th procedure. Future studies should correlate the learning curve with efficacy.
