Rectal cancer pelvic recurrence: imaging patterns and key concepts to guide treatment planning

Abstract

For rectal cancer, MRI plays an important role in assessing extramural tumor spread and informs surgical planning. The contemporary standardized management of rectal cancer with total mesorectal excision guided by imaging-based risk stratification has dramatically improved patient outcomes. Colonoscopy and CT are utilized in surveillance after surgery to detect intraluminal and extramural recurrence, respectively; however, local recurrence of rectal cancer remains a challenge because postoperative changes such as fat necrosis and fibrosis can resemble tumor recurrence; additionally, mucinous adenocarcinoma recurrence may mimic fluid collection or abscess on CT. MRI and ¹⁸F-FDG PET are problem-resolving modalities for equivocal imaging findings on CT. Treatment options for recurrent rectal cancer include pelvic exenteration to achieve radical (R0 resection) resection and intraoperative radiation therapy. After pathologic diagnosis of recurrence, imaging plays an essential role for evaluating the feasibility and approach of salvage surgery. Patterns of recurrence can be divided into axial/central, anterior, lateral, and posterior. Some lateral and posterior recurrence patterns especially in patients with neurogenic pain are associated with perineural invasion. Cross-sectional imaging, especially MRI and ¹⁸F-FDG PET, permit direct visualization of perineural spread, and contribute to determining the extent of resection. Multidisciplinary discussion is essential for treatment planning of locally recurrent rectal cancer. This review article illustrates surveillance strategy after initial surgery, imaging patterns of rectal cancer recurrence based on anatomic classification, highlights imaging findings of perineural spread on each modality, and discusses how resectability and contemporary surgical approaches are determined based on imaging findings.

Graphical Abstract

Introduction

Locally recurrent rectal cancer, defined as an intrapelvic recurrence following primary surgical resection (with or without distant metastases), remains a critical challenge for multidisciplinary rectal cancer teams. Previously, local recurrence was seen in over 20% of cases [1]. Due to the implementation of total mesorectal excision (TME) to achieve negative margins (R0 resection) and neoadjuvant therapy protocols, which employ local staging with MRI, the incidence of local recurrence has dramatically improved over the past two decades [2]; however, local recurrence still occurs in 2.4 to 10.0% of cases [3, 4]. Clinical manifestations of local recurrence are frequently non-specific and may include pain, bleeding, and urinary or rectal obstruction; nonetheless, most recurrences are identified on surveillance cross-sectional imaging [5]. Predictors of locally recurrent rectal cancer include surgical factors (i.e., abdominoperineal resection, higher residual tumor (R) classification [R1/2 is worse than R0], intraoperative perforation), primary tumor factors (i.e., low rectal tumors, advanced T-stage, positive circumferential resection margin [CRM], lymphovascular invasion, extramural venous invasion, poor tumor differentiation, and extracapsular invasion of nodal metastasis), and postoperative CEA levels [6-8].

Despite the poor outcomes in patients with locally recurrent rectal cancer (5-year survival: 9%), curative resection, which may require pelvic exenteration depending on tumor extent, improves the prognosis (5-year survival 57%) [9]. Additionally, when feasible, the inclusion of intraoperative radiation treatment (IORT) can be beneficial to further improve outcomes [10]. Appropriate patient selection for curative resection after local recurrence often requires multimodality imaging with computed tomography (CT), magnetic resonance imaging (MRI), ¹⁸F-fluoro-deoxyglucose (¹⁸F-FDG)-positron emission tomography CT (PET/CT), and/or ¹⁸F-FDG PET/MRI [11].

Cross-sectional imaging plays an important role not only in determining resectability but also in directing surgical approach [11, 12]. Locally recurrent rectal cancer is frequently linked to distant recurrence; indeed, over 50% of local recurrences are associated with distant metastasis at the time of diagnosis [13]. In determining potential surgical candidates, CT and/or ¹⁸F-FDG PET/CT examinations identify distant metastasis while liver MRI, preferably with a hepatocyte specific agent, is most accurate for diagnosing hepatic metastases [14]. After excluding unresectable distant metastasis, surgical resectability of the local pelvic recurrence is best determined based on pelvic MRI given its high-contrast resolution, which allows for detailed evaluation of radial spread of tumor. ¹⁸F-FDG PET/MRI is evolving as a robust modality to delineate perineural invasion.

This review article illustrates imaging findings of pelvic rectal cancer recurrence in different patterns based on anatomic classification, highlights imaging findings of perineural spread on each modality, and discusses how resectability and the contemporary surgical approaches are determined based on imaging findings.

Surveillance strategy

The most common clinical manifestations of locally recurrent rectal cancer are rectal bleeding, change in bowel habits, and pain. Pain as a presenting symptom should raise suspicion for pelvic organ, nerve, or bone involvement. However, over 23–30% of patients with local recurrence are asymptomatic and diagnosed at scheduled surveillance [15]. Such routine surveillance, with a combination of history, physical exam, carcinoembryonic antigen (CEA), colonoscopy, and contrast-enhanced abdominopelvic CT, is recommended for all stage II-III rectal cancer patients who have undergone a curative oncologic resection [16, 17], with most local recurrences occurring within 2 years after initial resection (median: 1.3–1.9 years) [3, 18]. CT surveillance in these patients is recommended every 6–12 months for 5 years [19]. For patients with stage I disease, routine imaging surveillance is not recommended; however, surveillance CT imaging is performed based on suspicious symptoms of recurrence because the recurrence rate is relatively low (4%) [20]. Colonoscopy is recommended at 1 and 3 years after surgery, thereafter every 5 years unless an advanced lesion is revealed [21].

Colonoscopy and CT are complementary in diagnosing local recurrence. Colonoscopy enables early detection of anastomotic recurrence by demonstrating mucosal abnormalities; however, it is unable to detect extraluminal recurrence at the anastomosis resection margins, nodal metastasis, and distant metastasis. Surveillance imaging with chest and abdominopelvic CT enables detection of extraluminal recurrence not revealed on colonoscopy. CT often demonstrates postoperative and post radiation changes including fat stranding and soft tissue thickening due to edema, inflammation, fibrosis, and fat necrosis [22], which can cause confounding imaging findings mimicking local recurrence (Fig. 1). Radiologists reviewing postoperative imaging need to be familiar with these findings to prevent false-positive results and be sensitive to changes in these findings on subsequent studies, which may enable early detection of recurrence. In case of equivocal imaging findings, MRI, ¹⁸F-FDG PET/CT, or biopsy should be considered depending on the local availability and preference.

Fig. 1.

Recurrence of rectal cancer. A 55-year-old male underwent abdominoperineal resection with colostomy with intraoperative radiation for rectal cancer 1 month ago. The inflammatory change observed in the fat tissue in the omental flap after 1 month (a: asterisk) is reduced after 6 months (b: asterisk). An irregular soft tissue mass is not changed (arrows). After an additional 12 months, CT and PET/CT demonstrate a newly enlarged enhanced soft tissue associated with FDG uptake in the same location (c and d: arrows). The recurrent tumor was proven with a CT-guided biopsy

Surveillance strategy can be altered based on the presumed probability of recurrence based on imaging findings and respective institutions and countries. For example, if CT findings are likely to be benign (e.g., representing inflammation or postsurgical change), the CT surveillance interval may be shortened (< 6–12 months) depending on the physician and patient’s concerns. This approach is simple and non-invasive, but potentially contains of risk of further progression if recurrence is present. Also, if follow-up CT remains inconclusive, further investigation will still be required. MRI, ¹⁸F-FDG PET/CT, and ¹⁸F-FDG PET/MRI serve as adjuncts to routine surveillance CT and colonoscopy for problem-solving and subsequent decision making, particularly if there is increased clinical suspicion based on clinical-stage, patient symptoms, and/or CEA levels.

MRI can also further elucidate indeterminate findings identified at routine surveillance because of its higher diagnostic performance than CT. The combination of T1, T2, and enhancement properties are helpful for characterizing a mass or for suggesting alternative benign conditions [23]. Diffusion-weighted imaging can help highlight recurrences that may be initially overlooked because fibrosis demonstrates low signal intensity while residual tumor shows higher signal intensity [24]. However, mucinous carcinoma should be recognized as a pitfall because of its high intensity on T2-weighted images and poor enhancement, which can occasionally simulate a complex fluid collection or abscess. Mucinous recurrence generally appears as a heterogeneously hyperintense mass on T2-weighted imaging with faint islands of enhancing tumor often observed within the nonenhancing mucin (Fig. 2) [25].

Fig. 2.

Recurrence of mucinous adenocarcinoma. A 75-year-old male underwent proctocolectomy with end ileostomy for anastomotic recurrence of rectal cancer 7 years ago. He developed urine incontinence and pelvic mass without FDG uptake 1 year ago. The mass with high signal intensity on T2WI (a and b: arrows) associated with rim enhancement (c: arrows) are observed adjacent to the bladder tethered to the pelvic wall which mimics a part of the bladder

Further evaluation of the indeterminate findings with ¹⁸F-FDG PET/CT is often useful in differentiating post-treatment change vs. recurrence) (Fig. 1). Additionally, ¹⁸F-FDG PET/CT permits detection of nodal and distance metastases. Recurrence commonly shows higher uptake of ¹⁸F-FDG than fibrosis or fat necrosis because ¹⁸F-FDG uptake is correlated with glucose metabolization; however, there are known pitfalls [26, 27]. For example, since as shown in pretreatment assessment by ¹⁸F-FDG PET/CT, ¹⁸F-FDG uptake is determined not only by glucose metabolic activity but also cell density; mucinous and signet cell carcinomas, which are associated with low cellularity and small tumors, may produce false negative results [28, 29]. Conversely, active inflammation may cause false-positive results due to high ¹⁸F-FDG uptake. When compared to MRI alone, ¹⁸F-FDG PET/MRI increases reader confidence and decreases the number of equivocal cases despite comparable diagnostic performance [30]. ¹⁸F-FDG PET/MRI can provide detailed tumor spread especially perineural invasion with high-contrast resolution for surgical planning.

Classification of locally recurrent rectal cancer

There are several anatomy-based classification systems for of locally recurrent rectal cancer which help determine both treatment plan and prognosis [12, 31]. The anatomic location is classified into the anterior, sacral, and pelvic sides in the Mayo Clinic classification [32], or the axial/central including anastomotic, anterior, lateral, and posterior in the Memorial Sloan-Kettering classification (Figs. 3 and 4; Table 1) [11, 33-35]. Lateral and posterior invasion generally demonstrate worse outcomes than axial recurrence because it reduces the likelihood of surgical resection with negative margins [31, 36, 37]. In addition to the anatomic category on the axial plane, some classifications use craniocaudal extent with the peritoneal reflection as a boundary [38], as well as symptoms and fixation to adjacent anatomy [32]. Among these classifications, the Memorial Sloan-Kettering classification has been most frequently used in clinical studies [31]. Local recurrence except for intraluminal anastomotic recurrence is not visible on colonoscopy. Lateral and posterior recurrences may be associated with perineural invasion.

Fig. 3.

Lateral recurrence. A 60-year-old male underwent low anterior resection for rectal cancer 6 years ago. Rising CEA level raised clinical concern for recurrence and chest CT was concerning for bilateral lung metastases. He complained of pain in the anal region and right leg with a sciatic distribution. Extensive nodular high signal intensity masses suggesting mucinous tumor is demonstrated on the right pelvic sidewall on fat-suppressed T2-weighted image (a: arrows). Additionally, tumor spread surrounds the seminal vesicles and bladder base is clearly observed on diffusion-weighted image (arrowheads)

Fig. 4.

Posterior recurrence. A 45-year-old female who underwent low anterior resection 3 years ago became symptomatic with severe rectal pain. On CT, the right S3 sacral foramen is replaced with soft tissue density (arrow) which is in continuity with the recurrent tumor (a: asterisk). Enhancement within the foramen (b: arrow) and the presacral tumor with cortical irregularity of the sacrum (b: asterisk) is also seen on fat-suppressed postcontrast T1-weighted image

Table 1.

Sloane Kettering classification of locally recurrent rectal cancer [33]

Category	Prevalence	Site
Axial	13–37%	Anastomotic site, residual mesorectal or perirectal soft tissue, and perineum
Anterior	16–30%	Bladder, ureter, prostate, seminal vesicle, uterus, and vagina
Lateral	18–25%	bone, muscle, and soft tissue of pelvis sidewall
Posterior	10–41%	Sacrum, presacral fascia, sacral root sheath

Axial/central recurrence occurs in the anastomotic site, residual mesorectal, or perirectal soft tissue, or perineum (following abdominoperineal resection). Anastomotic recurrence is categorized into intraluminal, extramural, and combined, with extramural not being directly visible on colonoscopy but only on cross-sectional imaging [39]. Anastomotic recurrence accounts for 13–37% of locally recurrent rectal cancers [34, 35].

Anterior recurrence accounts for 16–30% of locally recurrent rectal cancer [34, 35]. Anterior recurrence involves genitourinary structures including the bladder, ureter, prostate, seminal vesicle, and uterus, and vagina. Lateral recurrence involves pelvic sidewall structures such as pelvic bones, muscles, and soft tissue, accounting for 18–25% of locally recurrent rectal cancers [34, 35]. Sidewall involvement may extend to involve anterior and posterior structures. One study reported no patients with tumor invasion to the sciatic nerve, greater sciatic foramen, or lateral pelvic wall survived over 5 years [36]. Occasionally lateral “recurrence” is actually untreated residual disease (e.g., a lateral pelvic lymph node metastasis not identified pre-surgically and subsequently not resected or treated). Posterior recurrence, including the involvement of the sacrum, presacral fascia, and sacral root sheaths, accounts for 10–41% of locally recurrent rectal cancer [34, 35]. The presacral involvement is possible in case of a recurrent tumor contact with the sacrum even without obvious sacral bone involvement. After curative surgery, only 10% of patients having posterior recurrence associated with sacral invasion show a 5-year survival [36].

Perineural spread

Perineural invasion is defined as histologic evidence of cancer invasion of a nerve (Fig. 5). Pathological perineural invasion in surgical specimens is hypothesized as an indicator of an aggressive tumor and the cause of local recurrence [40, 41]. The reported frequency of pathological perineural invasion is approximately 20.6% in rectal cancer and is recognized as a poor prognostic feature with a hazard ratio of 1.85–2.07 in overall survival [42]; therefore, similar to basic TNM staging and vessel invasion, pathological perineural invasion may be taken into account in stratifying therapeutic approaches [43]. Despite the fact that MR neurography is increasingly used to evaluate the lumbosacral plexus in patients with known or suspected lumbosacral plexopathy [44], there is limited literature on the imaging findings of perineural invasion although lateral and posterior local rectal cancer recurrence can be associated with perineural invasion of the sciatic nerve and L5-S1 [45, 46]. The identification of perineural invasion is greatly facilitated by understanding its common pattern and appearance at MR imaging.

Fig. 5.

Histopathological findings of perineural invasion. Invasive cancer foci (C) invade the nerves (N) and the investing perineurium (P) as shown on Hematoxylin–Eosin stein

The pelvic nervous system is categorized into somatic and autonomic nervous systems. The lumbosacral plexus is a network of somatic nerves supplying motor and sensory innervation to the pelvic organs and the lower extremities. The spinal nerves converge to form the lumbosacral plexus, which then diverges to form the following terminal nerves: the sciatic, femoral, obturator, superior, and inferior gluteal, iliohypogastric, ilioinguinal, genitofemoral, lateral femoral cutaneous, and pudendal [44]. The inferior hypogastric plexus (a.k.a., pelvic plexus) belongs to the autonomic nervous system in the pelvis. The inferior hypogastric plexus continues to the superior hypogastric plexus and sacral splanchnic nerves (sympathetic component) as well as pelvic splanchnic nerves arising at level S2-4 (parasympathetic component). The inferior hypogastric plexus provides autonomic visceral innervation and is distributed diffusely around the rectum, bladder, and vagina or prostate (Fig. 6) [47, 48]. During a routine TME, the hypogastric nerve trunks are spared, but peripheral branches will be removed with the mesorectum. Cancer cells can spread along these neural networks in both anterograde and retrograde directions.

Fig. 6.

Pelvic nerve anatomy. The sacral spinal nerves contribute to the lumbosacral plexus which then divides to form the terminal peripheral nerves (e.g., sciatic nerve). The inferior hypogastric plexus provides autonomic visceral innervation connected by the superior hypogastric plexus, sacral, and pelvic splanchnic nerves, and distributes diffusely around the rectum, bladder, and vagina or prostate

Imaging findings of perineural spread

The appearance of a normal peripheral nerve is that of a cord-like structure with a fascicular pattern, smaller diameter than the accompanying artery, isointense to skeletal muscle on T1-weighted and iso- to minimally hyperintense on T2-weighted fat-suppressed images, without enhancement on post-gadolinium images, and well outlined by perineural fat (Fig. 7) [44, 49]. Normal peripheral nerves are symmetrical and uniform in size and signal intensity and taper distally. The fat pad sign (loss of normal perineural fat) and the muscle sign (muscle denervation) have been described in perineural spread of tumor in head and neck malignancy and are accompanied by an abnormal appearance of the affected nerve (Table 2) [50]. The fat pad sign can also be helpful in diagnosing perineural tumor spread in rectal cancer, particularly as nerves travel through the transverse or sacral foramens (Figs. 4 and 8). On CT, perineural spread of malignancy results in the perineural fat density being replaced with enhanced soft tissue density. On MRI, high intensity around the nerve indicates preservation of the perineural fat on T1-weighted images and non-fat-suppressed T2-weighted images. Fat-suppressed sequences are suboptimal for the detection of the fat pad sign due to the low signal contrast with the surrounding tissue on this sequence (i.e., fat tissue, bone marrow). While it is difficult to demonstrate perineural invasion on CT, the superior contrast resolution of MRI allows for better depiction of the involved nerves, distinguishing them from the uninvolved ones (Figs. 9 and 10). Typically, an involved nerve is irregularly enlarged, shows high-intensity on T2-weighted and diffusion-weighted images, and enhances homogeneously or heterogeneously on post-gadolinium images (Fig. 9). In our experience, ¹⁸F-FDG-PET/MRI is a robust modality to demonstrate perineural tumor spread in recurrent rectal cancer owing to the functional benefits and high sensitivity in detecting tumor spread with ¹⁸F-FDG-PET and the anatomic benefits related to the high-contrast resolution of MRI to delineate peripheral nerve anatomy (Fig. 11) [51]. In case of suspected perineural invasion due to clinical symptoms or suspicious imaging findings on surveillance CT or MR, additional examinations (i.e., MRI using a dedicated lumbosacral protocol (MR neurography) and ¹⁸F-FDG-PET/MRI) may be employed to further investigate extent of bone and/or nerve involvement. Comparison with the opposite side is paramount in image interpretation of perineural tumor invasion because it is generally asymmetric.

Fig. 7.

Normal peripheral nerve on MRI. The bundle structures originated from the right L1-S2 nerve root, which shows iso-signal intensity on T1-weighted image (a), slightly high signal intensity on T2-weighted image (b), and no enhancement on post-gadolinium image (c), are entering to the right ischial tuberosity with forming the right sciatic nerve (arrows). The right S3 nerve root surrounded by perineural fat demonstrates the similar intensity (arrowheads)

Table 2.

Imaging findings of perineural spread [44; 49]

Modality	Nerve findings	Innervation muscle
CT	Irregularly enlarged nerve	Muscular atrophy and fat deposition in chronic phase
	Perineural fat replaced with soft tissue around sacral foramen
MRI	Irregularly enlarged nerve	Areas of hyperintensity on T2-weighted images (indicative of edema) (< 1 month)
	T2 intermediate to high intensity	Areas of hyperintensity on T2- (indicative of edema) and T1-weighted images (indicative of fatty infiltration) in subacute phase (1–3 monthes)
	DWI hyperintensity	Areas of hyperintensity on T1-weighted images (indicative of fatty infiltration) and reduced muscle volume (indicative of atrophy) (> 3 month)
	Enhancement on post-gadolinium image
	Perineural fat replaced with soft tissue around sacral foramen
18F-FDG-PET	Tubular or linear FDG uptake	No or mild FDG uptake

Fig. 8.

Fat pad sign and denervation in the perineural invasion. Fat density in the right sacral foramen of the S1 (a: arrow) and S2 (a: arrowhead) is replaced with soft tissue density. Despite no abnormality in the right piriformis muscle on CT (b: circle), there is mild asymmetric infiltrating high intensity within the muscle on the T1-weighted image (c: circle) compatible with fatty infiltration. Corresponding feathery asymmetric high intensity on fat-suppressed T2-weighted image (d: circle) suggests edema. These combined findings are consistent with subacute findings of denervation in the S1 and S2 innervated piriformis

Fig. 9.

Retrograde Perineural invasion from lumbosacral plexus to the spinal nerve. A 61-year-old male was treated with neoadjuvant chemoradiotherapy followed by lower anterior resection two years ago and noticed progressive numbness in his lower extremities. T2-weighted images demonstrates a tubular structure (a-c: arrows) in continuity with the recurrent tumor (a-b: asterisks). Similar imaging findings are more clearly observed in the diffusion-weighted (d-f) and post-gadolinium images (g-i)

Fig. 10.

Comparison of MRI and 18F-FDG-PET/MRI in the conspicuity of perineural invasion. A 63-year-old female who underwent neoadjuvant chemoradiotherapy followed by low anterior resection 4 years ago for rectal cancer complained of pelvic pain and disordered bowel function. The right sciatic nerve (a: arrow) and the S3 nerve root of (b, c: arrows) adjacent to the recurrence tumor (a and d: asterisks) show marked enhancement compared to the opposite side. Abnormal increased FDG uptake is clearly observed in the right sciatic nerve (d: arrow) and right S3 nerve root in the sacral foramen (e, f: arrows). Additionally, 18F-FDG PET/MRI demonstrates FDG uptake in the cauda equina (e: arrowhead)

Fig. 11.

Surgical planning to lateral recurrence. A 50-year-old male underwent abdominal perineal resection followed by neoadjuvant chemotherapy for rectal cancer (T3N1M0) 4 years ago. A nodal mass on the left internal iliac region was observed on CT and elevated CEA is revealed despite being asymptomatic. An irregular, poorly defined mass along the left pelvic sidewall (a, b: asterisks) with involvement of the wall of the bladder (a, b: black arrows) and the obturator internus muscle (b: white arrow). Based upon the imaging findings, en bloc resection including the left pelvic sidewall, partial cystectomy, left ureterectomy, and prostatectomy is performed with intraoperative radiotherapy. The residual bladder was sutured and a left ureteroneo-cystostomy was constructed. An omental pedicle flap was used to fill the pelvic void. Postoperative CT shows no residual tumor in the pelvis and the omental pedicle flap filling the dead space in the pelvis (c: asterisk)

In addition to the imaging changes of the nerve itself, secondary muscular findings of nerve involvement can also develop in the innervated territory (muscle sign). For example, early on intramuscular edema may develop followed by eventual atrophic fatty degeneration in the chronic phase [52]. Spinal nerves at levels L5 and S1 are commonly affected by recurrent rectal cancer, and thus, these changes may be seen in the obturator internus and gluteus muscles (Table 3) [53]. It is difficult to assess edema in the muscle in the acute phase with CT, but it can show muscle atrophy and fatty infiltration in the chronic phase. On MRI, intramuscular edema is demonstrated as a feathery hyperintensity throughout the muscle on T2-weighted images and fatty degeneration on T1-weighted images (Table 2).

Table 3.

Muscle and innervating nerve roots [53]

Muscle	L5	S1	S2	S3	S4
Obturator internus	*	*
Gluteus minimus	*	*
Gluteus medius	*	*
Gluteus maximus	*	*
Piriformis	*		*
Levator ani			*	*	*

The asterisks mean innervation between muscles and nerve roots

Contemporary surgical approaches to pelvic recurrence

Curative treatment for locally recurrent rectal cancer is exenterative en bloc resection of the tumor with part or all of the invaded organs to achieve sufficient tumor-free margins (R0). Although pelvic exenteration of recurrent tumor is associated with early and late postoperative complications including hemorrhagic, infectious, and obstructive complications in the abdomen and pelvis (37–100%) and a relatively higher postoperative mortality rate (0–25%), such radical surgery is performed in tertiary referral institutions in an attempt to achieve cure (median survival: 8–38 months; 5-year survival: 0–37%) [54].

Pelvic exenteration is anatomically classified into three types: anterior pelvic exenteration which is the removal of the rectum, reproductive organs, and bladder; posterior pelvic exenteration which is the removal of the rectum; and total pelvic exenteration is the elimination of all pelvic organs [55]. Owing to the ability to reconstruct after sacrectomy [56], extended pelvic exenteration combined with concurrent sacrectomy is indicated for cases with suspected sacral invasion in posterior and total pelvic exenteration [57]. Individualized approaches are considered to preserve intact organs from tumor invasion based on imaging findings. The resulting potential space is reconstructed by a vascularized tissue flap such as omental and muscle flaps (Figs. 11 and 12) [58].

Fig. 12.

Surgical approach to perineural spread. A 36-year-old female underwent a laparoscopic low anterior resection with colorectal anastomosis 18 months ago for rectal cancer (stage III; pT4, N2) and completed adjuvant chemotherapy. She complained of right buttock pain radiating down her leg. A presacral mass compatible with local recurrence (a: asterisk) continues into the right S4 foramen (a: white arrow). Fusiform nodule indicates intramuscular perineural invasion (a: black arrow). On the basis of the imaging findings, an abdominal perineal resection (APR) with colostomy and a partial sacrectomy were performed as well as intraoperative radiotherapy. The sacral defect was closed with a right vertical rectus abdominis myocutaneous flap. Postoperative CT demonstrates the resected right side of the sacrum,vertical rectus abdominis myocutaneous flap (b: asterisk), and fluid collection (b: arrow) in the muscle defect

R0 resection is the most critical factor affecting prognosis in rectal cancer recurrence. Invasion of adjacent organs also diminishes the likelihood of R0 resection. Distant metastasis is also a poor prognostic factor and contraindication for salvage surgery because there is no difference in median survival whether or not a local recurrence is resected [6]. Generally, the indications for surgery in locally recurrent rectal cancer are the following: (1) no unresectable distant metastasis; (2) feasibility of an R0 resection or an R1 resection with minimal residual disease amenable to IORT in preoperative evaluation; and (3) physically and mentally fit to undergo invasive surgery [59].

The next step after excluding unresectable distant metastasis is resectability judgment and surgical planning. MRI offering high-contrast resolution is generally employed for this purpose. Although the indications are usually determined by a multidisciplinary approach in each institution, the following are generally considered contraindications at present: (i) invasion of the spine above S1, (ii) multifocal pelvic lesions or extensive pelvic sidewall involvement, (iii) invasion into sciatic notch, and (iv) encasement of common or external iliac vessels. Owing to improved surgical techniques, aggressive surgical approaches have become feasible in tertiary referral high-volume centers [60]. Even in combination with IORT, complete surgical resection is essential as 5-year survival is degraded by residual tumor (36–71% in R0 [no gross residual tumor with microscopic negative margin], 11–33% in R1 [removal of all macroscopic tumor but positive microscopic margin], and 13–25% in R2 resection [gross residual tumor]) [61]. ¹⁸F-FDG PET/MRI may be particularly useful in patients undergoing resection of locally recurrent rectal cancer, with one study estimating management changes in over 25% of patients [62]. Understanding the appearance and pattern of neural invasion in locally recurrent rectal cancer is essential for the evaluation of nerve involvement to accomplish R0 resection (Fig. 12). Because neurogenic pain is one of the symptoms of locally recurrent rectal cancer and is highly suspicious of nerve involvement, ¹⁸F-FDG PET/MRI may be considered to evaluate perineural spread.

Summary

Locally recurrent rectal cancer remains a major challenge despite improvement in patient outcomes due to wider adoption of TME and neoadjuvant and adjuvant therapies. CT is routinely employed for regular surveillance, and additional evaluation with ¹⁸F-FDG PET-CT, MRI, or ¹⁸F-FDG PET-MRI may be used for further evaluating indeterminate imaging findings on CT. More advanced imaging may help determine whether to continue surveillance or to perform an image-guided biopsy. The pattern of locally recurrent rectal cancer is anatomically classified into axial/central, anterior, lateral, and posterior recurrence with lateral and posterior recurrence associated with worse outcomes. Cross-sectional imaging may delineate perineural invasion, which is a form of recurrent tumor spread along and around the nerve(s), especially in lateral and posterior recurrence. The imaging findings of perineural invasion are nerve abnormalities (enlarged irregular nerve, high intensity on T2- and diffusion-weighted image, enhancement, and perineural fat replacement with tumor) and muscle change caused by denervation. Because ¹⁸F-FDG PET-MRI is an accurate modality to diagnose perineural tumor spread [11, 30, 62], it should be considered in patients with neurogenic pain and clinically suspected perineural invasion. Pelvic exenteration linked to the contemporary surgical planning is essential to achieve negative margins by en bloc resection of the tumor and invaded organs. Imaging plays two important roles in evaluating resectability and surgical planning. One is to exclude unresectable distant metastasis. The other is to determine the feasibility of an R0/R1 resection, a determination which is primarily based on pelvic MRI and ¹⁸F-FDG PET-MRI images. The indication and surgical planning should be judged through multidisciplinary discussions.