Carotid interposition in patients with head and neck tumors: clinical experience of 13 cases reconstructed with a great saphenous vein autograft

Abstract

Background

Carotid interposition graft (CIG) surgery in the setting of head & neck cancer (HNC) is a rare procedure with a limited number of cases described in the literature.

Aims/Objectives

To assess the outcomes of the surgery at Helsinki University Hospital.

Materials and methods

Patients who underwent CIG in a head and neck tumor surgery were retrospectively analyzed over 15 years. Overall-survival (OS) was calculated until 1 May 2020. The primary-outcome was to measure the 30-day OS, postoperative stroke rate, and other complications. The secondary-outcome was to measure 1-, 2-, and 5-year OS.

Results

Thirteen patients were identified, 11 with HNC and two with Shamblin III Carotid Body Tumors. The great saphenous vein was used for all vascular reconstructions, and shunting was routinely performed. The 30-day stroke incidence was nil. Two graft-blowouts were encountered, one of which lead to death and the other was successfully managed. For HNC patients, the locoregional recurrence-rate was 36%. The 5-year OS was 46.2%.

Conclusion and significance

CIG in HNC setting can achieve oncologic-control with an acceptable rate of complications. Routine shunting, heparinization, and elevating blood-pressure during closure seem to be safe protocols to maintain cerebral-circulation perioperatively. A moderate graft-blowout risk should be considered.

Chinese Abstract

背景：颈动脉介入移植（CIG）手术对于头颈癌（HNC）是一种罕见的手术, 文献描述的病例数量是有限的。

目的：评估赫尔辛基大学医院的手术结果。

材料和方法：回顾性分析了15年内在头颈部肿瘤手术中接受 CIG 的患者, 计算了直至2020 年 5月1日的总生存期 (OS)。主要结果是测量 30 天 总生存期、术后卒中率和其他并发症。次要结果是测量1年、2年和 5年总生存期 。

结果：确认了13名患者, 其中11名患有头颈癌, 2名患有 Shamblin III 颈动脉体肿瘤。大隐静脉用于所有血管重建, 并按常规进行分流。 30天卒中发病率为零。遇到了两次移植爆喷, 其中一次导致死亡, 另一次被成功控制。对于头颈癌患者, 局部区域复发率为36%。 5年总生存期为46.2%。

结论与意义：针对头颈癌的 颈动脉介入移植可实现肿瘤控制, 且并发症发生率可接受。关闭期间的常规分流、肝素化和血压升高似乎是围手术期维持脑循环的安全方案。应考虑中度移植爆喷风险。

Keywords:

• Carotid interposition graft
• carotid artery reconstruction
• head and neck cancer
• surgery

Introduction

Adequate local control of head and neck tumors that involve the carotid artery requires radical surgery [1,2]. Carotid ligation without reconstruction has a high risk of stroke [2,3]. Thus, many studies have examined the feasibility of reconstructing the resected carotid artery with an interposition graft [4]. This technique has proved safe, with a lower risk of stroke, better local control, and palliative benefit compared with surgery without vascular reconstruction [4–7]. However, the number of cases described in the literature is still limited. It also remains controversial what tests should be done to estimate the risk of carotid ligation [4–6,8], what kind of materials should be used for vascular reconstruction [4,8–11], and whether the quality of life can be satisfactory after such a radical procedure [5].

This study reports our experience of 11 patients with head and neck cancer (HNC) and two patients with carotid body tumors (CBT) operated in the Helsinki University Hospital, Helsinki, Finland, between 2004 and 2019. We hypothesized that a carotid interposition graft (CIG) is a safe technique, and this audit was performed to assess this.

Materials and methods

This retrospective cohort study was conducted at the Helsinki University Hospital, Helsinki, Finland, a tertiary health care facility with a referral region covering a population of 1.8 million. Patients who underwent CIG in a head and neck tumor operation between April 2004 and May 2019 were retrospectively identified and analyzed. The patients were retrieved from the electronic archives using relevant operative codes for common carotid artery (CCA) and internal carotid artery (ICA) interposition graft. Patients with a cervical tumor or a tumor involving the ipsilateral neck were identified. Data regarding pre-, peri- and postoperative characteristics were retrieved, and overall survival (OS) was calculated until May 1, 2020. HNCs were classified following the 8^th edition TNM classification. The Adult Comorbidity Evaluation 27 (ACE-27) was used to assess preoperative comorbidities [12], and postoperative complications were evaluated with the Clavien-Dindo classification [13].

Statistical analysis was performed with the SPSS version 23.0 statistical software package (IBM Corporation, Armonk, NY, USA). Due to the small sample size, continuous variables were reported as median and range. Categorical variables were reported as counts and percentages. A Kaplan-Meier curve was plotted for the postoperative overall survival. We reported overall survival separately for patients with HNC and CBT.

The study's primary outcome was to measure the 30 day OS, postoperative stroke rate, and other postoperative complications. The secondary outcome was to measure 1-, 2-, and 5-year overall survival rates.

The study was done following the ethical standards of the institutional and national research committees and in line with the 1964 Helsinki declaration and its later amendments. According to the Finnish legislation, as this study was a retrospective chart review, no informed consent was needed. Institutional research permission was granted for the study protocol (HUS/307/2019 June 19, 2019).

Surgical technique

Our surgical team for tumors encapsulating the carotid artery routinely involves vascular surgeons, plastic surgeons, head and neck surgeons, and maxillofacial surgeons. Extension of the disease defines surgical exposure. When the tumor is adjacent (less than approximately 2 cm) to the skull base, or the tumor extension necessitates, we consider utilizing a previously published midline mandibulotomy technique to access the ICA medially [1]. In other cases, we use a lateral incision with or without resection of the styloid process (Figure 1). In these cases, the extended lateral exposure includes nasal intubation or tracheostomy and subluxation of the mandible when necessary. Our threshold to perform midline mandibulotomy was lower earlier in the series. However, after gaining experience with extended lateral mobilization, especially resection of the styloid process to better expose the skull base, we chose to operate on more patients without a mandibulotomy. Although lateral mandibulotomy techniques are also used in selected cases, all mandibulotomies in this series were anterior midline. In our experience, the proximal part of the mandible and the temporomandibular joint do restrict the exposure in lateral mandibulotomies. The benefit of a midline mandibulotomy is the possibility to approach the distal ICA medially from the nasopharynx above the tonsil.

Figure 1. A lateral incision with resection of the styloid process (A) distal anastomosis site highlighted near the skull base (B).

In this study, a GSV autologous graft was used in all cases. No vascular prostheses were used due to the risk of bacterial contamination. After encountering a blowout incident at the anastomosis site, we started using free muscular fascia to strengthen the vascular anastomosis, especially in the radiated field (Figure 2). This technique was copied from our experiences in the surgical management of aortic infections [14]. Following several fistula complications, we later became more meticulous in minimizing that risk, encasing the CIG with a pedicled or free flap. In more cranial reconstructions, the external carotid artery was permanently sacrificed.

Figure 2. Use of free muscular fascia to strengthen the vascular anastomosis. Anastomoses being made (A) Anastomoses completed (B).

We did not conduct pre-and/or intraoperative tests to assess the potential risk of perioperative stroke. Our protocol includes routine shunting with a Pruitt–Inahara shunt, heparinization, and elevating blood pressure to above 170 mmHg during closure. The ischemia time is limited as both anastomoses are almost finalized before shunt removal, and the final 2–3 stitches are sutured after flushing. It would be technically possible to close one anastomosis first and perform the shunting via the CIG while suturing the other anastomosis. However, we believe this is unnecessary. Permanent antiplatelet therapy was prescribed postoperatively for all patients.

Results

The study included ten males (77%), and three females (23%), with a median age of 59 years (range, 21–69). Nine patients (69%) had a previous smoking history, and five patients (38%) had a history of alcohol abuse. Twelve patients (92%) underwent surgery with curative intent. The study comprised 11 patients with HNC, of which 10 had squamous cell carcinoma (SCC), and one had adenoid cystic carcinoma (ACC). The p16 tumor suppressor gene correlates with a good prognosis in head and neck SCC. In 2014 we started examining the p16 immunohistochemical staining characteristics in oropharyngeal cancers, and three out of five tested cases were p16-positive (Table 1).

Table 1. Pre- and intraoperative characteristics of the study sample.

Patient	Year of Surgery	Age (Sex)	ACE-27^[Citation12]	Histology	Primary Tumor	Classification	Previous Treatments	Interposition level	Approach	Flap used	Follow-up
1	2004	36 (M)	0	SCC	Oral Cavity	T2N0M0 (R)	RT & S	CCA-ICA	MM	DCIA	6 y, 7 m, 3 d
2	2005	48 (F)	2	SCC	Oropharynx	T2N3aM0 (P)	No	CCA-ICA	MM	Pectoralis	0 y, 3 m, 11 d
3	2006	21 (M)	0	CBT	Carotid Body	Shamblin 3 (P)	Embolization	CCA-ICA	MM	No	14 y, 3 m, 6 d
4	2012	28 (F)	0	CBT	Carotid Body	Shamblin 3 (P)	Embolization	ICA-ICA	MM	No	7 y, 6 m, 8 d
5*	2014	54 (M)	0	SCC^a	Oropharynx	T4N2M0 (R)	CRT & S	CCA-ICA	ELI	Pectoralis	6 y, 0 m, 15 d
6	2014	61 (M)	1	SCC	Unknown	T0N3bM0 (P)	No	CCA-ICA	ELI + SR	Pectoralis	1 y, 1 m, 12 d
7**	2015	59 (F)	3	SCC	Oropharynx	T2N3bM1(P)	No	CCA-ICA	ELI + SR	LD	0 y, 2 m, 16 d
8***	2017	69 (M)	9	SCC^a	Oropharynx	T2N1M0 (R)	CRT & S	CCA-ICA	ELI + SR	LD	2 y, 10 m, 11d
9	2018	64 (M)	2	SCC	Hypopharynx	T2N2bM0 (R)	RT	CCA-CCA	ELI	ALT + Pectoralis	0 y, 5 m, 2 d
10	2018	68 (M)	1	SCC	Oropharynx	T4aN2bM0 (R)	CRT & S	CCA-ICA	ELI + SR	LD	0 y, 0 m, 23 d
11****	2018	51 (M)	1	SCC	Oral Cavity	T3N0M0 (P)	RT & S	CCA-CCA	ELI	No	2 y, 2 m, 18 d
12*	2018	63 (M)	0	ACC	SMG	T2N0M0 (R)	RT & S	ICA-ICA	ELI	ALT	0 y, 0 m, 10 d
13	2018	59 (M)	1	SCC^a	Unknown	T0N3M0 (P)	No	CCA-ICA	ELI	Pectoralis	0 y, 3 m, 19 d

M: male; F: female; SCC: squamous cell carcinoma; CBT: carotid body tumor; ACC: adenoid cystic carcinoma; SMG: submandibular gland; RT: radiotherapy; CRT: chemoradiotherapy; S: surgery; R: recurrence; P: primary; CCA: common carotid artery: ICA: internal carotid artery; MM: midline mandibulotomy; ELI: extended lateral incision; SR: styloid resection; LD: latissimus dorsi; DCIA: deep circumflex iliac artery; ALT: anterolateral thigh; Y: year; M: month ; D: day.

^aP16 positive; *A blow-out complicated the CIG; **operated on with palliative intent; ***intravascular stent prosthesis was implanted postoperatively following flap necrosis and graft exposure; ****operated on an emergency basis after a carotid blow-out.

In addition, our series included two patients who underwent surgery for a large primary Shamblin III CBT. Their main symptom was a neck mass, and one of them complained of cardiac palpitations. Both cases had SDHD gene mutations and negative urine metanephrine expression (Table 1).

Seven patients were operated on for primary tumors, including the CBTs (54%). In one patient, the tumor had primarily been peeled off the carotid wall, which resulted in a blowout in the intensive care unit (ICU) and required cardiopulmonary resuscitation followed by immediate surgery, including CIG. The remaining six patients (46%) were operated on for recurrent tumors. From the whole patient cohort, seven patients had a history of chemoradiotherapy (CRT) to the head and neck site (one patient with a primary tumor and six patients with a recurrent tumor). Neck dissection was previously performed on the ipsilateral tumor side in five patients. To achieve proper distal control of the ICA, a midline mandibulotomy with lateral rotation of the mandible was performed in four patients (31%), and the styloid process was resected through a lateral incision in four patients (31%); otherwise, the procedure was conducted through an extended lateral neck incision. In 10 patients with HNC, a muscular or myocutaneous flap was used to cover the CIG, most often the pectoralis flap (5 patients − 50%) (Table 1). Twelve (92%) patients had one cranial nerve sacrificed, and 10 (77%) patients required the sacrifice of 2 or more: the vagus nerve in 11 (85%), the accessory nerve in 10 (77%), and the hypoglossal nerve in 10 patients (77%). The median operative time was 550 min (range, 283–705).

Table 2 displays the perioperative management of the study cohort. The median postoperative hospitalization time was 15 days (range, 8–48). Seven patients (54%) required postoperative care in the ICU for a median of 4 days (range, 2–6). The median time before decannulation for patients with a tracheostomy was nine days (range, 2–14). Three patients (23%) required a permanent tracheostomy.

Table 2. Perioperative management of the study sample.

Variables
	n	%
Aspiration protection
NGT	5	38
PEG	8	62
Perioperative tracheostomy	6	46
Permanent tracheostomy	3	23
Need for ICU follow up	7	54
	Median	Range
ICU follow up time (days)	4	2–6
Time before decannulation (days)	9	2–14
Hospital stay (days)	15	8–48

NGT: nasogastric tube; PEG: percutaneous endoscopic gastrostomy; ICU: intensive care unit.

Within 30 days postoperatively, none of the patients suffered a stroke, and vascular graft donor site complications were not encountered. One patient (8%) died of a graft blowout on the 10th postoperative day (POD), while another died on the 23rd POD due to ischemic heart disease. Within 60 days of the primary surgery, all patients developed at least one postoperative complication, and six of them (46%) required additional surgery. A second blowout was encountered during a fistula revision surgery on the 38th POD. The affected distal anastomosis part of the CIG was excised, and the graft defect was covered with a GSV graft. The patient survived for more than five years. In addition, three patients experienced wound complications, and two patients developed partial flap necrosis. The vascular graft was exposed in one of these patients, and an intravascular stent prosthesis was implanted to limit the risk of a blowout. The most common complications following nerve damage were dysphagia (12 patients − 92%), dysarthria (10 patients − 77%), and neck and/or shoulder stiffness (8 patients − 61%) (Table 3).

Table 3. Postoperative complications within 30 days postoperatively following the Clavien-Dindo Classification^[13].

Grade	n	%
Grade I
Dysphagia	12	92
Dysarthria	10	77
Neck/Shoulder stiffness	8	61
Jaw stiffness	6	46
Facial expression compromise	5	38
Wound dehiscence	1	8
Grade II
Pneumonia	5	38
Grade IIIb
Fistula formation	4	31
Wound dehiscence	2	15
Flap vascular complications	2	15
Graft blowout	1	8
Flap donor site complication	1	8
Grade V
Death	1	8

The figures in the table are not mutually exclusive.

Postoperatively, the institutional multidisciplinary tumor board advised the two CBT patients to receive radiotherapy (RT) and three HNC patients to receive CRT. Locoregional recurrence was detected in four (36%) HNC patients after a median postoperative interval of 98 days (range, 50–200). The recurrence was treated with CRT in two patients, and two received palliative treatment. Bone metastasis in the lumbar spine was detected in one patient with CBT after four years of the surgery and was treated with RT. The 1-year OS was 53.8% for the whole cohort, and both 2- and 5-year OS were 46.2%. Both CBT patients were alive and recurrence-free at the end of their follow-up at 7.5 and 14.3 y. The 1-year OS for HNC was 45.5%, and both 2- and 5-year OS were 36.4% (Figure 3).

Figure 3. Kaplan–Meier plot for overall survival (OS) of patients, carotid body tumors (CBT) reported separately.

Discussion

Main finding

In this study, 13 patients were managed with a carotid interposition GSV graft in a multidisciplinary approach for tumors involving the carotid artery. No postoperative strokes were encountered, but a moderate graft blowout rate (2 patients 15%). The 5-year OS was 46%, the locoregional recurrence rate for HNC patients was 36%.

Perioperative stroke

We do not conduct pre-and/or intraoperative tests to assess the potential risk of perioperative stroke due to tests' unreliability, and our 30-day stroke rate was nil. In a larger sample of 51 patients, Mourad et al. conducted routine shunting on 51 patients and experienced stroke in 2 patients (3.9%) [6]. He et al. combined preoperative temporary balloon occlusion testing (TBO) with shunting and experienced 1 stroke in 10 patients [10]. However, a literature review of 357 tumor cases with CIG detected no relationship between preoperative testing, use of shunts, and postoperative stroke [4]. TBO test shows false negativity and would need sufficient occlusion times and a hypotensive challenge to improve its sensitivity [2]. Local anesthesia, somatosensory-evoked potentials (SEP), motor evoked potentials (MEP), electroencephalography (EEG), near-infrared spectroscopy (NIRS), and transcranial Doppler (TCD) are techniques used to estimate the need for intraoperative shunting during carotid surgery [15]. Local anesthesia, TCD, and NIRS are routinely used in our institution. However, local anesthesia is impossible due to the extensive dissection needed in these cases; the TCD head frame was considered too cumbersome. NIRS is feasible but still unreliable. Besides, its use is logistically challenging in our institution, as the H&N tumor operations are physically performed in a different hospital building than routine carotid surgery. As none of the methods is perfect, we consider routine shunting, heparinization, elevating the blood pressure during closure to be safe protocols to maintain cerebral circulation without testing cerebral perfusion.

Other surgical options to CIG, such as permanent ligation of the carotid artery, involve a 17–79% risk of stroke [5]. In addition, ligation does not exclude the possibility of a blowout. Peeling the tumor off the carotid wall has been reported to increase the risk of blowouts due to the excision of the protective vascular adventitia [6]; one such patient was included in our series. However, it is difficult to define when the procedure should be called peeling. Besides, tumor recurrence rates following peeling can also reach 50% due to compromised resection [6]. Preoperative covered stenting can be a protective measure before tumor peeling and would facilitate reconstruction even within the skull base [16]. However, larger studies are still needed to support the safety of this method.

Vein graft and shunting

In this study, a GSV autologous graft was used in all cases. Some authors report on polytetrafluoroethylene (PTFE), suggesting its resistance to damage by fistulas and its ability to maintain patency in the challenge of postoperative RT [5,10]. In our series, we experienced several fistula cases that healed without a blowout. Other authors opt for using the superficial femoral artery (SFA), arguing for its better resistance to infection [9,11]. Yokoyama et al. advocate the superficial femoral vein (SFV) as a graft material; the main argument is the relatively larger diameter that would make the anastomosis technically easier, thus limiting the ischemic time [17]. In our institution, the femoral vein is frequently used for reconstructions in the infected aortic field [14]. In our experience, it is too large for the ICA but could be considered in the CCA reconstructions. Our primary choice, GSV, is the most widely used graft in the literature, with a limited incidence of stroke and blowout complications [8,18–20]. It is rarely associated with vascular donor site complications, requires no donor-site reconstruction, causes almost no neural damage, and its size is suitable for the ICA. However, a meticulous follow-up of these patients is needed to improve understanding of the graft material compliance in the long term, especially if exposed to radiation and oral contamination.

Mortality and morbidity

We encountered two postoperative blowouts, one was corrected successfully, and the patient survived for more than five years, while the other led to acute hematemesis and death on the 10th POD. After this blowout, we developed the technique and introduced strengthening the vascular anastomoses with free muscular fascia, especially when the neck was previously dissected and/or radiated (Figure 2). We detected a late bone metastasis in one of the 2 CBT patients during follow-up. For the remaining 11 cancer patients, the locoregional recurrence rate was (36%). At five years follow-up, the OS was 46.2%. According to previous articles, the resection and reconstruction of the involved carotid artery with a CIG achieve better local tumor control, reduce the risk of stroke to 3.7% while improving the 5-year survival rates up to 52.4% [4]. The procedure also prevents distressing death modes, whether from a carotid blowout, an ulcerating neck mass, or airway compression, thus achieving a palliative benefit [1,4,5,7,8]. Our results are comparable to previous reports on this procedure and add to the consensus that it can achieve an oncologic benefit with acceptable morbidity.

Limitations

This is a single institution series of a rare medical condition in a retrospective approach, with related limitations. While our number of cases is comparable to other studies published on the same topic, the sample is still small and subject to bias. Additionally, significant diversity in the underlying pathology could influence survival rates and make it difficult to extrapolate results to daily practice.

Conclusions

CIG in the head and neck cancer setting can achieve oncologic control benefits with an acceptable rate of complications and better survival rates than other treatment modalities. Routine shunting, heparinization, and elevating blood pressure during closure seem to be safe protocols to maintain cerebral circulation without the need for preoperative testing of cerebral perfusion. A moderate graft blowout risk should be considered.

Acknowledgment

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Disclosure statement

No potential conflict of interest was reported by the author(s).