Keywords
Anorectal malformation - distal cologram - MRI
Introduction
Anorectal malformations (ARM) comprise a wide spectrum of diseases, which can involve
the distal anus and rectum as well as the urinary and genital tracts. Incidence is
approximately 1:5,000 live births with a slight male predominance.[1] ARM since long has been classified by the Wingspread classification of 1984 based
on puborectalis (PR) sling. More recent Krickenbeck classification, which is based
mainly on the presence or absence of fistulas, their type and location, as well as
the position of the rectal pouch is widely accepted today.[2],[3]
ARM is commonly associated with other developmental anomalies. Most common association
is with genitourinary system, while vertebral, spinal, skeletal, cardiovascular, and
gastrointestinal anomalies as well as syndromic associations are also observed. This
is important because the overall prognosis and the quality of life is guided by the
severity of the associated anomaly. Once the diagnosis of the specific defect is established,
the functional prognosis can be rapidly predicted, which is vital to avoid raising
false expectations in the parents. Factors such as the status of the spine, sacrum,
and perineal musculature influence the prognosis. Patients with a hypo-developed sacrum
and sphincter muscle complex (SMC) are much more likely to be incontinent.
The most commonly used operative procedures for treatment of ARMs are posterior sagittal
anorectoplasty (PSARP) and laparoscopicor laparotomic anorectoplasty (LAARP). These
operative procedures require specific anatomical information for which clinical examination
is not enough. Imaging plays a key role in establishing and evaluating the real position
of the rectal pouch; the presence or absence of rectourogenital fistulas and their
location; the grade of development of SMC.
Accurate assessment of ARM has long been considered difficult. Invertography was the
earliest imaging technique used but its estimation of rectal pouch is highly inaccurate.[4] Ultrasonography (US) is a good diagnostic modality for imaging pelvic structures,
distal rectal pouch, and internal fistulas but does not show the SMC directly. It
has been advocated as a screening method for associated spinal and genitourinary anomalies.[3] Limited studies have been done on infracoccygealtransperineal US but its role is
still not established.[5] Voiding cystourethrography (VCUG) has also been used in patients with urinary tract
abnormalities at US or clinical suspicion of rectourinary fistulas (meconium in urine)
and identification of vesicoureteric reflux (VUR). High-pressure distal colostography
is currently considered the most efficient imaging technique for demonstrating level
of rectal pouch and fistulas; however, no information regarding development of muscles
of continence and associated anomalies can be obtained.[6] Computed tomography (CT) allows direct visualization of SMC but is poor in soft
tissue characterization. High dose of ionizing radiation limits its use in children.
Few studies have been done to evaluate ARM with magnetic resonance imaging (MRI) but
MRI has no defined status in the imaging protocol of ARM. Recent advances in MRI technology
allow fast and high-resolution imaging and visualization of small pediatric pelvic
structures in good detail. An attempt should be made to reassess its role in evaluation
of the level of rectal pouch and type of ARM, fistulas, SMC development, and the possibility
to determine associated anomalies, especially of spinal cord, spine and the urogenital
system in a single examination.
The present study was attempted to further clarify the utility of MRI in ARM and its
role in imaging protocol.
Materials and Methods
A cross-sectional observational study was conducted after approval of institutional
review board and ethics committee. Informed consent was obtained from the parents.
Study was done for a period of 1.5 years from October 2013 to March 2015. Thirty pediatric
patients of age less than 14 years (19 boys and 11 girls) with suspected ARM on clinical
evaluation underwent MRI. Patients who had been previously undergone reparative surgery
for ARM or had any contraindications to MRI or sedation were excluded. Distal cologram
(DC) was done in 26/30 patients who underwent colostomy. Colostomy was not done in
four female patients with anovestibular fistula with no colonic outflow obstruction.
US was done for urogenital anomalies in all the 30 patients. MRI interpretation was
done prior to DC to prevent any bias.
Magnetic resonance imaging scanning method and parameters
MRI was done with a 1.5-T magnet (PhilipsAchieva). The patients were placed in a supine
position with the pelvis positioned within the phased array body coil. The patients
were administered adequate sedation. No contrast instillation was done through the
perineal orifices. Multiplanar T1-weighted and T2-weighted images were obtained in
all patients [Table 1]. Fat suppression was used in young infants with thin fat planes where routine T2-weighted
images were not sufficient.
Table 1
Magnetic resonance imaging (MRI) protocol
|
Sequence
|
Slice width (mm)
|
TR/TE (ms)
|
Flip angle (°)
|
FOV (mm)
|
Imaging plane
|
Anatomical region
|
|
*Fat suppression was done wherever T2W images were not adequate, mostly in young infants,
TSE: Turbo spin echo, FOV: Field of view, TR: Repetition time, TE: Echo time
|
|
T2W TSE
|
3
|
4500/120
|
90
|
300
|
Coronal
|
Pelvis, spine (from D10), kidneys
|
|
T2W TSE
|
3
|
4500/125
|
90
|
150
|
Sagittal
|
Pelvis , lumbosacral spine
|
|
T2W TSE
|
3
|
4500/120
|
90
|
200
|
Axial
|
Pelvis, lumbosacral spine
|
|
T1W TSE
|
3
|
760/15
|
90
|
200
|
Axial
|
Pelvis, lumbosacral spine
|
|
T2WTSE SPAIR*
|
4
|
2500/70
|
90
|
150
|
Sagittal
|
Pelvis , lumbosacral spine
|
The imaging protocol was modified in view of patient’s individual requirements wherever
needed. Approximate scan time was 8–10 min.
Image interpretation
Images were evaluated for: (1) level of rectal pouch in relation to pelvic floor (2)
presence or absence and type of fistula (3) subjective developmental state of SMC
and levator ani (4) length and width of PR muscle (5) length and width of external
anal sphincter (EAS) (6) lumbosacral spine and spinal cord anomalies (7) renal and
urinary tract anomalies (8) genital tract anomalies. T2-weighted images in both axial
and sagittal planes were used in correlation for detection of fistula. The distinction
between bowel and fistula was made according to presence or absence of layered bowel
wall respectively. Normal bowel was identified by the T2 hyperintense mucosa, while
the wall of fistula was homogenous with no central hyperintense mucosa [Figure 1]. The development of levator ani, PR, and EAS were assessed subjectively as good,
fair, and poor. A well-developed levator ani was seen clearly as a sling-like structure
on coronal images supporting the rectal ampulla. PR muscle was seen as a triangular
muscular structure on axial images surrounding the rectum with apex posteriorly. EAS
was seen as a posterior curved band like structure on sagittal images, with fibers
extending in parasagittal images and as an oval structure symmetrically surrounding
rectum/anal canal on axial images. Deviation from the normal well-defined appearance
was evaluated as “fair” and when the muscle fibers were poorly visualized, they were
graded as “poor.”Thickness and length of the PR and EAS were also objectively measured
and following indices were calculated. Pubococcygeal (PC) distance was distance from
inferior border of pubic symphysis to sacrococcygeal joint. “I” distance was half
of the distance between the inner border of ischial tuberosities. Total width of PR
and EAS was considered the sum of both left and right muscle width of rectum or anal
canal. Both were measured along 3 o’clock and 9 o’clock at mid-level of anal canal.
Figure 1 (A-C): Level of rectal pouch (*) as identified by hyperintense mucosa lying above, at or
below pelvic floor (arrow) on T2-weighted sagittal image. (A) High type. (B) Intermediate
type. (C) Low type. Also note that rectum is opening into the vestibule in (C) (rectovestibular
fistula)
Relative width of PR (RWPR) = (Total width of PR)/(half of “I” distance).
Relative length of PR (RLPR) =(Length of PR)/(PC distance).
Relative width of EAS (RWEAS) = (Total width of EAS)/(half of “I” distance).
Relative length of EAS (RLEAS) = (Length of EAS)/(PC distance).
The MRI findings were correlated with the preoperative observations. At surgery, the
level of rectal pouch in relation to the pelvic floor was ascertained. Anatomy of
any fistula present was noted as per Krickenbeck classification. The development of
SMC was graded according to subjective assessment of muscle thickness by the pediatric
surgeon preoperatively as good, fair, or poor.
Statistical analysis
The sensitivity, specificity, and accuracy of MRI in defining the level of rectal
pouch, fistula detection, and development of SMC were calculated. The sensitivity,
specificity, and accuracy of DC in defining the level of rectal pouch and fistula
detection were also obtained. Chi-square test was used to calculate the significance
of MRI findings.
Results
There were 19 male and 11 female patients. Most of the patients (27/30) underwent
MRI at <3 years of age. Age distribution was <6months (N = 16), 6 months to 3 years (N = 11), 3–7 years (N = 1) and 7–14 years (N = 2). There were 18/19 males with no anal opening and 1/19 male with ectopic anus
with rectal atresia. Four perineal openings were present in a female with H-type fistula.
Surgery confirmed that among 19 males, level of rectal pouch was above, at and below
the pelvic floor in nine, four, and six patients respectively. Among 11 females, level
of rectal pouch was above, at and below the pelvic floor in one, zero, and eight patients
respectively, while two were cloaca. The results of MRI and DC with respect to level
of rectal pouch are summarized in [Table 2]. Overall accuracy of MRI and DC to detect the exact level of rectal pouch including
cloacal malformation was 93.33% and 76.9% respectively. DC could not define the level
of rectal pouch in two patients due to lack of bony landmarks consequent to anomalies
like pubic diastasis and partial sacral agenesis. Other causes of inaccuracy were
over distension or under distension of rectal pouch. MRI misinterpreted the level
in two patients due to inadequate visualization of bowel mucosa and poor image quality.
Table 2
Level of rectal pouch in relation to pelvic floor as assessed on surgery
|
Level of rectal pouch
|
Surgery
|
Correctly diagnosed on MRI
|
Correctly diagnosed on DC
|
|
MRI: Magnetic resonance imaging, DC: Distal cologram
|
|
Above pelvic floor
|
9
|
8
|
8
|
|
At pelvic floor
|
5
|
5
|
3
|
|
Below pelvic floor
|
14
|
13
|
7 (out of 10)
|
|
Cloaca
|
2
|
2
|
2
|
|
Total
|
30
|
28
|
20
|
On surgery, 21 (70.00%) cases had fistula including 10 males and 11 females. Fistula
was absent in nine patients; all male, with imperforate anus. The results of MRI and
DC in detecting presence or absence of a fistula are given in [Table 3]. MRI and DC could correctly identify presence or absence of fistula in 76.6% and
76.9% cases respectively. The sensitivity and specificity of MRI and DC in detecting
presence of fistula were 76.19%and 77.78% and 70.59% and 88.89% respectively.
Table 3
Presence of fistula as assessed in surgery
|
Fistula
|
Surgery
|
Correctly detected on MRI
|
Correctly detected on DC
|
|
MRI: Magnetic resonance Imaging, DC: Distal cologram
|
|
Present
|
21
|
16
|
12
|
|
Absent
|
9
|
7
|
8
|
|
Total
|
30
|
23
|
20
|
Type of fistula was categorized as per Krickenbeck classification [Figure 2], [Figure 3], [Figure 4], [Figure 5]. The accuracy of MRI and DC in correct diagnosis of anatomical type of fistula was
calculated as given in [Table 4]. MRI and DC correctly identified the anatomy of fistula in 76% and 65% cases respectively.
DC was more specific than MRI in detecting presence of fistula but less accurate in
anatomical characterization [Figure 6], [Figure 7], [Figure 8].
Table 4
Fistula characterization as per Krickenbeck classification on surgery and number of
correct diagnosis on magnetic resonance imaging and distal cologram
|
Krickenbeck classification
|
Surgery
|
MRI
|
DC
|
|
MRI: Magnetic resonance imaging, DC: Distal cologram
|
|
Rectoperineal
|
1
|
1
|
-
|
|
Bulbar recto-urethral
|
2
|
2
|
1
|
|
Prostatic recto-urethral
|
6
|
4
|
3
|
|
Bladder neck fistula
|
4
|
2
|
2
|
|
Vestibular fistula
|
4
|
3
|
1
|
|
Cloacal malformation
|
2
|
2
|
0
|
|
No fistula
|
9
|
7
|
9
|
|
H-type
|
1
|
1
|
0
|
|
Rectovaginal
|
1
|
1
|
1
|
|
Anal stenosis
|
0
|
0
|
0
|
|
Total
|
30
|
23/30 (76%)
|
17/26 (65%)
|
Figure 2: Fistula (white arrow) with urinary tract (black arrow) can be distinguished from
the rectum (*) by absence of T2 hyperintense mucosa. (A) Bladder neck fistula. (B)
Prostatic rectourethral. (C) Bulbar rectourethral
Figure 3: H-type fistula is seen extending from the rectum at point A to the vestibule at point
B. Anal opening is located at point C
Figure 4: Cloacal anomaly. The urethra (white arrow), vagina with hydrocolpos (*) and rectum
(black arrow) are seen to open into a short common channel
Figure 5: Blind ending rectal pouch (*) with no fistula is seen having a clear fat interface
with the urinary tract (white arrow) and atretic anal canal (black arrow)
Figure 6 (A and B): 2-month-old female with single perineal opening at birth. (A) Distal cologram (DC)
shows rectovaginal fistula. (B) T2-weighted sagittal MRI image shows convergence of
rectal pouch, vagina and urethra (arrow), clearly depicting cloacal anatomy
Figure 7 (A and B): 18-month-old male child. (A) Distal cologram (DC) shows high rectourethral fistula.
(B) T-weighted sagittal MRI image clearly depicting anatomy as hypointense fistulous
tract (arrow) leading from the rectal pouch to the lower prostatic part of urethra
Figure 8 (A-D): 10-month-old child. (A) Distal cologram (DC) shows anterior beaking (arrow) in the
rectal pouch suspicious of fistula, although no further opacification was achieved
due to reflux from the colostomy site. (B) T2-weighted MRI image shows hypointense
tract (arrow) from the rectal pouch to the urethra. (C) Axial T2-weighted image shows
the rectal pouch (black arrow) close to the prostate (white arrow). Note the central
hyperintense mucosa allowing differentiation from fistula. (D) Axial image caudal
to it leaves no doubt to the anatomy as the hypointense fistula (black arrow), lacking
a central bright mucosa as seen in the rectal pouch, is directly visualised entering
the prostatic parenchyma (white arrow) to join the prostatic urethra
Lumbosacral spine anomaly was present in 9/30 (33.3%) patients, and included partial
sacral agenesis (N = 5), block vertebra (N = 3), hemivertebra (N = 1). Spinal cord anomalies were seen in two patients including lipomyelomeningocele
with tethered cord and presacral lipoma in one patient [Figure 9] and filar lipoma with tethered cord in another. It was found that 50% of spinal
anomalies were present in high type, 33.33% in intermediate type and 16% in low type.
Renal and urinary tract anomalies were present in 12 patients (40% of cases) [Figure 10]A and [Figure 10]B. The most common anomaly was vesicoureteral reflux (VUR) 6/30 (20%) followed by
renal agenesis 5/30 (16.67%). Others were hydronephrosis (N = 3), atrophic kidney (N = 2), and horseshoe kidney (N = 1). Renal and urinary tract anomalies were most common in high type of ARM (88.9%
of anomalies), followed by intermediate (40%) andlow type (7%). Genital anomaly was
present in 3/30 (10%) patients, including both cloacal anomalies. One of the cloacal
anomalies had associated hydrocolpos and other had bicornuate uterus. Hydrocolpos
was also present in another patient of low type of ARM.
Figure 9: Currarino's triad comprising anorectal malformation, partial sacral agenesis with
presacral mass (*). Spinal cord is tethered (black arrow) to the presacral lipoma
Figure 10 (A and B): Complex urinary anomaly. (A) Horseshoe kidney is seen with hydroureteronephrosis.
(B) Large thick walled neurogenic bladder with Foley's catheter in situ is seen in
the sagittal image
The development of levator ani and PR were assessed subjectively on MRI as good, fair,
and poor and compared with the level of ARM as found on surgery. It was found that
higher the ARM, poorer the levator ani and PR development as seen on MRI. The association
was statistically significant with P- value of 0.008 and 0.023 respectively. A correlation
was made between appearances of levator ani, PR, and EAS on MRI [Figure 11]A and [Figure 11]B and overall assessment of SMC on surgery subjectively as good, fair, and poor.
The correlation was found to be significant for each with P- values of 0.003, 0.019,
and 0.016 respectively [Table 5]. Therefore, MRI can accurately predict development of SMC in ARM.
Figure 11 (A and B): Muscle development. (A) Levator ani is graded as poor development on subjective assessment.
(B) Levator ani is graded as good development on subjective assessment
Table 5
Comparative study of subjective assessment of muscle development on magnetic resonance
imaging with the sphincter muscle complex assessment on surgery
|
MRI appearance
|
SMC as assessed on surgery
|
Total
|
P
|
|
Poor
|
Fair
|
Good
|
|
EAS: External anal sphincter
|
|
Levator ani
|
|
|
|
|
|
|
Poor
|
4
|
3
|
0
|
7
|
0.003
|
|
Fair
|
4
|
3
|
2
|
9
|
|
|
Good
|
1
|
2
|
11
|
14
|
|
|
PR
|
|
|
|
|
|
|
Poor
|
5
|
5
|
0
|
10
|
0.019
|
|
Fair
|
4
|
2
|
4
|
10
|
|
|
Good
|
0
|
1
|
9
|
10
|
|
|
EAS
|
|
|
|
|
|
|
Poor
|
6
|
4
|
1
|
11
|
0.016
|
|
Fair
|
3
|
4
|
7
|
14
|
|
|
Good
|
0
|
0
|
5
|
5
|
|
|
Total
|
9
|
8
|
13
|
30
|
|
MRI was also used to determine the development of PR and EAS using objective indices.
Relative width of PR and EAS were compared with the development of SMC on surgery
and the results were not statistically significant (P = 0.3394; 0.1297 respectively).
Discussion
The study was done to evaluate pediatric patients of ARM with MRI. There was male
preponderance; M:F ratio being 1.72:1 (19/11 patients), which is in concordance with
the incidence data in literature.[3]
The most common presenting complaint was nonpassage of meconium at birth (63.33%),
and was seen in 94.73% of males. Second most common complaint was leakage of feces
in urine. In females, the presenting complaint was much more variable.
This study revealed high type ARM as the most common type in males which is in concordance
with previous literature.[3],[7] Most commonly seen ARM variant in females was low type (72.72%). This is consistent
with the study by Hashmi MA et al.[8] where analysis of 130 female patients of ARM over 10 years revealed a prevalence
of low type lesions in 72% patients.
DC was done in 26 out of 30 cases because four patients were passing feces through
the ectopic anal/vestibular/normal opening and in them colostomy was not done. The
overall accuracy of DC in determining level of rectal pouch was 76.9%.
Overall accuracy of MRI in determining the level of rectal pouch was 93.33%. It is
similar to one of the studies by Nievelstein et al.[9] in year 2002, where MRI could correctly depict the levels in 96% cases. The accuracy
of DC in determining level of ARM is limited byim proper distension and difficult
visualization of bony landmarks in patients with lumbosacral anomalies and scoliosis.
On the other hand, the rectal wall and mucosa as well as pelvic floor is directly
visualized on MRI and hence, the assessment of rectal pouch level is more accurate.
Rectovestibular fistula was the most common fistula in females and rectourethral was
most common in males. Hashmi et al.[8] and Peña [7] also report the same.
In the present study, it was found that fistula was present in 70% of cases. Nievelstein
et al.[10] in year 2002 observed fistula on MRI in 62.5% patients. The present study reveals
that MRI is 76.19% sensitive and 77.78% specific in fistula detection. MRI and DC
correctly identified the fistula anatomy in 76% and 65% cases respectively. Similar
results were reported by Thomeer et al.[11] in year 2015 found that MRI and DC could correctly predict anatomy of fistula in
88% and 61% cases respectively.
The present study shows that MRI is more sensitive but less specific in fistula detection
when compared with DC. This may be attributed to the fact that in DC, contrast is
instilled with an augmented pressure to identify a fistula which increases the specificity
of this investigation in detecting fistulas. However, the anatomical type of fistula
is more accurately depicted by MRI than DC due to direct visualization of pelvic structures.
Lumbosacral spine and spinal cord anomalies have been reported in 41% to 60% of cases
with ARM on MRI.[5],[7],[10],[12],[13] In the present study, frequency of associated lumbosacral spine anomalies were 33.33%
and majority of them were present in high type (50%). Genitourinary anomalies were
commonly associated, most frequent being VUR followed by renal agenesis. Previous
literature also showed similar results.[13],[14],[15],[16],[17],[18]
Development of SMC plays important role in restoration of the bowel function after
surgical correction of ARM. Few studies have been done to assess the PR and EAS on
MRI. The developmental state of SMC has been positively correlated in different types
of ARM.[19],[20] Patients with low anomaly are expected to have good development of skeletal muscle
mass.
In our study, the subjective appearance of levator ani, PR, and EAS on MRI relative
to the SMC assessment on surgery were statistically significant (P < 0.05). Objective measurement of the SMC on MRI was however, not found to correlate
well with the preoperative assessment. The diagnostic criteria for the poor developmental
state of muscles were defined by Shah et al.[19] as RWPR <0.18 and RWEAS <0.15. In the present study, these cut-off values were not
found to be significant with P-values of 0.339 and 0.129 respectively [Table 6].
Table 6
Range of the relative values for muscle development
|
RLPR
|
RWPR
|
RLEAS
|
RWEAS
|
|
RLPR: Relative length of puborectalis, RWPR: Relative width of puborectalis, RLEAS:
Relative length of external anal sphincter, RWEAS: Relative width of external anal
sphincter
|
|
Poor
|
0-0.4
|
0-0.13
|
0
|
0-0.29
|
|
Fair
|
0.49-0.55
|
0-0.37
|
0-0.27
|
0.16-0.21
|
|
Good
|
0.31-0.63
|
0-0.48
|
0.31-0.56
|
0.17-0.32
|
The refinements in surgical management of ARM call for a more precise preoperative
local anatomical evaluation.[11],[21] The benefits over single stage over staged repair have long been advocated as better
continence and easier dissection in neonate due to virgin anatomical planes. The poor
social acceptance of colostomy is also an issue. The hazards of single stage repair
are mainly due to intraoperative injury to urinary tract because of lack of anatomical
knowledge otherwise provided by DC.[22] MRI has the potential to circumvent this problem in a noninvasive manner and equip
the surgeon for a single stage repair.
Postoperatively, MRI can play a role in the evaluation of patients with persistent
fecal incontinence. MRI is undoubtedly the optimal imaging modality for assessing
these patients who are underconsideration for reoperation. Axial images are generally
best for assessing thesiting of the pulled-through rectum and also for evaluating
the puborectalis and EAS.[23] Anterior misplacement of the neorectum in the externalanal sphincter, and lateral
misplacement of the neorectum in the puborectalis muscle, are the most common errors
observed.[24]
Many technical innovations have been used in the recent past to extract more information
from MRI. Jarboe et al. in 2012 reported a Combined 3D rotational fluoroscopic-MRI cloacagram procedure
by instilling contrast through catheters placed into mucous fistula, cloacal channel,
bladder, vagina, rectum in complex pelvic malformations.[25] Recent studies have successfully attempted 3D MRI reconstruction, 3D MRI genitography,
and MRI-guided LAARP.[21]
There were many limitations in our study. There was limited number of participants.
We did not perform gadolinium contrast installation through the colostomy or fistula
site for the MRI. We did not follow-up these patients after surgery and did not study
the difference in postoperative outcome of these patients with those in which MRI
was not done. A study with greater number of participants with long-term follow-up
of postoperative outcome would better validate the role of MRI in these patients.
Conclusion
The preoperative imaging evaluation of ARM presently involves DC, fistulogram, voiding
cystourethrogram, ultrasound of the kidneys or bladder and spine, and MR imaging of
the spine with DC being the current gold standard for precise anatomy of distal rectum.
The present study has shown that MRI provides accurate answers for most of the preoperative
questions and scores over DC at many fronts. MRI can accurately evaluate the level
of rectal pouch by direct identification of pelvic floor and also the rectum by its
hyperintense mucosa, thereby eliminating indirect assessment on DC by hypothetical
radiographic lines. Pelvic musculature including LA, PR, and EAS can be directly assessed.
Fistula can also be directly visualized without the limitations of variation in pressure
by rectal distension. There is added advantage of simultaneous depiction of associated
spinal and genitourinary anomalies. MRI can accurately evaluate multiple facets of
the clinical problem in a single investigation. With advances in MRI technology and
newer innovations in MRI technique, it is expected to assume a more important role
in the preoperative evaluation of ARM.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms.
In the form the patient(s) has/have given his/her/their consent for his/her/their
images and other clinical information to be reported in the journal. The patients
understand that their names and initials will not be published and due efforts will
be made to conceal their identity, but anonymity cannot be guaranteed.
