Rectal Cancer

(1)

Rectal Cancer

The Influence of Surgical Technique on Morbidity, Mortality and Survival

Petrus Boström

Department of surgical and perioperative sciences Umeå 2020

(2)

This work is protected by the Swedish Copyright Legislation (Act 1960:729) Dissertation for PhD

ISBN: 978-91-7855-393-8 (print) ISBN: 978-91-7855-394-5 (pdf) Cover design by Marie-Therese Vinnars

Electronic version available at: http://umu.diva-portal.org/

Printed by: Cityprint i Norr AB Umeå, Sweden 2020

(3)

Dedication

To my grandparents, whose blood runs in my veins and into this book

Vad du ej klart kan säga, vet du ej med tanken ordet föds på mannens läppar det dunkelt sagda är det dunkelt tänkta¹

1Ursprungligen från Epilog vid magisterpromotionen i Lund, 22 juni 1820, av Esaias Tegnér. Återberättat för mig i ett av min morfars, Ingvar Aronsson, många tal, i detta fall på Grand Hotel i Lund, någon gång under min barndom.

(4)

(5)

Abstract

Even though rectal cancer management is the archetype of multimodality treatment, no modality increases the chances for cure more than surgical resection. During the last century, surgical methods have vastly improved and become more standardised, but controversies prevail regarding surgical decision- making. In addition, the introduction of total mesorectal excision surgery and neoadjuvant treatment improved the oncological results, but also increased the complications of surgery, most notably anastomotic leakage. The aim of this dissertation is to evaluate the short-term risks and long-term benefits of surgery for rectal cancer, allowing patient and surgeon to make more informed treatment decisions.

By cross-referencing the Swedish Colorectal Cancer Registry with the Prescribed Drugs Register, a cohort of all anterior resections in Sweden in 2007–10, performed on patients with increased cardiovascular risk, was defined. Operative charts were retrieved to evaluate the association between high tie and anastomotic leakage in a logistic regression model. A high tie was defined as either ligation of the inferior mesenteric artery, proximal to the take-off of the left colic artery, or as a combination of ligation of the left colic artery and the superior rectal artery. A low tie was defined as ligation of only the superior rectal artery, distal to the take-off of the left colic artery. After exclusion, 722 patients remained for analysis. Symptomatic leakage was found in 12.3% (41/334) versus 10.6%

(41/388) in high and low tie, respectively, yielding an insignificant association between tie level and leakage overall (odds ratio 1.05; 95% confidence interval 0.61–1.84). However, in a post hoc analysis, patients with a history of both manifest cardiovascular disease and American Society of Anesthesiologists score III–IV seemed to be at greater risk for leakage after high tie (odds ratio 3.66; 95%

confidence interval 1.04–12.85). For these patients, it might be wise to either abstain from anterior resection or at least try to mitigate the risks and be wary of signs of anastomotic leakage if sphincter-saving surgery is performed.

In order to yield a population-based estimate of anastomotic leakage and postoperative mortality after anterior resection, all patients who underwent anterior resection in the period 2007–16 were identified through the Swedish Colorectal Cancer Registry. A prolonged postoperative period of 90 days was used as outcome, and anastomotic leakage and reintervention due to leakage as exposure. Directed acyclic graphs were used to select possible confounders for the logistic regression analyses. Of 6,948 patients, 693 (10.0%) experienced anastomotic leakage and 294 (4.2%) underwent reintervention due to leakage.

The mortality rate was 1.5% among patients without leakage, and 3.9% for those who suffered from leakage, but only patients in need of a reintervention displayed

(8)

increased mortality when controlling for confounders (odds ratio 5.57; 95%

confidence interval 3.29–9.44). There was also evidence of a substantial mortality increase in elderly patients with leakage (p = 0.007). It seems as if anastomotic leakage after anterior resection is still rather common and, in cases of more severe leakage, associated with postoperative mortality, a risk that becomes more pronounced with increasing age.

To investigate the impact of tie level on oncological outcomes after rectal cancer surgery, we identified and followed until 2018 all stage I–III rectal cancer patients who underwent abdominal surgery with curative intention in 2007–

2014, using the Swedish Colorectal Cancer Registry. As the level of tie is not randomised, a propensity score matched cohort was created, and Cox regression analyses on both the matched and unmatched datasets were performed. Cancer- specific survival was the primary outcome, while both overall and relative survival, as well as recurrence and lymph node harvest, were secondary outcomes. Of the 8,287 patients who remained for analysis, 37% had high tie surgery. After propensity score matching, the 5-year cancer-specific survival rate was 86% overall and no association between tie level and any survival or recurrence measure was found (hazard ratio 0.92; 95% confidence interval 0.79–

1.07 for cancer-specific survival). Total lymph node harvest was, however, increased after high tie surgery (p < 0.01), but no differences were seen regarding the number of positive nodes (p = 0.72). All subgroup and interaction analyses were similarly insignificant. Hence, the level of tie was not associated with any patient-oriented oncological outcome. This should allow the patient’s anatomical configuration and the surgeon’s preferences to determine the level of tie.

A final aspect of anastomotic leakage, given that it is still a major clinical problem, is the need of a timely diagnosis before septic symptoms develop. We therefore created a population-based cohort from the Swedish Colorectal Cancer Registry and the Swedish Perioperative Registry, yielding 3,084 patients who underwent anastomotic colorectal surgery for cancer in 2014–2017. The aim was to evaluate the association between early postoperative pain, measured with the numerical rating scale, and leakage or reoperation due to leakage. Of the 189 patients who suffered from anastomotic leakage, 121 required a reoperation. Both the univariable and multivariable logistic regression analyses revealed a biological gradient of increased risk for both leakage (odds ratio 2.38, 95% confidence interval 1.44–3.93 for pain score 8–10 compared to 0–3) and the need for reoperation with increasing postoperative pain. Though the absolute risk increase for leakage is rather modest, increased postoperative pain should probably induce extra vigilance among those caring for patients, who have recently undergone anastomotic colorectal surgery.

(9)

In conclusion, it seems as if the choice of ligation level influences morbidity and mortality more than survival in rectal cancer surgery. Even though high ligation overall seems to be a safe method, it is probably best avoided in patients with significant cardiovascular disease. This dissertation highlights the fact that anastomotic leakage continues to pose a large clinical problem, as it is a strong driving force behind postoperative mortality, especially among the elderly, and only avoided with certainty through non-anastomotic surgery. In addition, our increased knowledge of risk factors has not significantly diminished its incidence, so it remains difficult to predict, even though increased pain could be a helpful clue in the postoperative setting. Future research that emphasises the risks and benefits of sphincter-saving surgery, optimal level of ligation and other potential risk factors for anastomotic leakage is needed.

(10)

Abbreviations

APE Abdominoperineal excision (of the rectum) ASA American society of anesthesiologists CI Confidence interval

CME Complete mesocolic excision CT Computer tomography DAG Directed acyclic graph

EMR Endoscopic mucosal resection ESD Endoscopic submucosal dissection

Gy Gray (SI unit of cumulative absorbed radiation) HR Hazard ratio

IMA Inferior mesenteric artery LCA Left colic artery

NPR The National Patient Register NRS Numerical rating scale

OR Odds ratio

PDR The Prescribed Drugs Register RCT Randomised controlled trial

SCRCR The Swedish Colorectal Cancer Registry SPOR The Swedish perioperative registry SRA Superior rectal artery

TEM Transanal endoscopic microsurgery TME Total mesorectal excision

TNM Tumour – Nodes – Metastases

(11)

Populärvetenskaplig sammanfattning

Bakgrund

Tjock- och ändtarmscancer är en den fjärde vanligaste cancerformen i världen och den tredje vanligaste orsaken till död i cancer. Naturalförloppet leder ofta just till döden, inte sällan i rikliga plågor. För ungefär 200 år sedan påbörjade dock en medicinsk, kirurgisk och vetenskaplig utveckling som sakta men säkert har ökat möjligheterna till bot och gjort kirurgi vid tillståndet allt säkrare. Men ända sedan man började operera ändtarmscancer via buksnitt har man diskuterat hur nära stora kroppspulsådern som man bör dela kärlen till tarmen. Ju mer centralt man delar kärlen, desto mer lymfkörtlar borde man teoretiskt få med sig i resektionen, men desto sämre kan blodförsörjningen till kvarvarande delen av tarmen vara.

Från och med 1980-talet har man börjat operera ändtarmscancer med en teknik som följer de naturliga skikten i buken och lilla bäckenet, vilka bildas redan under fostertiden. Denna teknik, tillsammans med strålning riktad mot ändtarmscancer, har kraftigt minskat risken för återfall i sjukdomen. En nackdel är dock att det samtidigt har ökat risken för läckage i den skarv som ibland anläggs i slutet av operationen, om man väljer att sammankoppla tarmen igen.

Samtidigt har vi med ökad kunskap om tumörbiologin kunnat avgöra vilka tumörer som kan tas bort endast lokalt och vilka som kan strålas enbart, varvid man kan undvika större kirurgi. Därmed kan naturligtvis komplikationer till kirurgi också undvikas. Särskilt läckage från tarmskarven kan leda till inflammerad bukhinna, blodförgiftning och död. Vi har dock i dag ökad kunskap om ytterligare riskfaktorer för denna komplikation, såsom manligt kön, övervikt, avancerade tumörer, lågt anlagda tarmskarvar, samsjuklighet och rökning.

Dessutom vet vi i dag att en avlastande stomi, d.v.s. en avledning av avföringen till en påse på magen medan ändtarmsskarven läker, minskar konsekvenserna vid läckage. Men än så länge är det några steg kvar innan vi kan lägga all risk och nytta för olika behandlingsalternativ i vågskålen och sedan välja den behandling som bäst balanserar dessa för varje individuell patient.

Denna avhandling designades därför för att utreda risker och nytta med att dela kärlen centralt – nära stora kroppspulsådern – eller mer distalt – närmre tarmen, samt för att utreda hur vanligt läckage i tarmskarven är i dag och hur ofta det leder till död efter kirurgi. Slutligen försökte vi utreda huruvida kraftig smärta efter operation ökar sannolikheten att patienten drabbas av läckage ifrån sin tarmskarv.

(12)

Material och metoder

Samtliga delarbeten använder det svenska kolorektalcancerregistret, där i stort sett all kirurgi för tjock- och ändtarmscancer registreras. I samband med diagnos, initiala röntgenundersökningar, operation, 30 dagar efter operation, vid tilläggsbehandling och vid långtidsuppföljning rapporterar sjukhusen till registret, vilket bidrar till 97% komplett överensstämmelse med journaldata för ändtarmscancer. Dessutom har i avhandlingen ytterligare svenska register använts, där läkemedelsregistret och patientregistret ingick i delarbete I, för att identifiera hjärt-kärlsjukdom, totalbefolkningsregistret i delarbete II, III och IV för att erhålla data om död, dödsorsaksregistret i delarbete III för att fånga cancer-specifik överlevnad och svenskt perioperativt register i delarbete IV för att framför allt få data om smärta efter operation. Dessutom har internationella Human Mortality Database använts i delarbete III, för att beräkna relativ överlevnad. Utöver registeruppgifter har vi i delarbete I inhämtat data direkt från patientjournaler.

Delarbete I studerade huruvida hög kärldelning (nära stora kroppspulsådern) oftare orsakade läckage från tarmskarven än låg kärldelning (nära tarmen).

Patienter med hjärt-kärlsjukdom, vilka därmed kanske inte har någon välfungerande ytterligare kärlförsörjning till tarmen utgör den grupps som borde ha störst nytta av en låg kärldelningsnivå, om nu kärldelningsnivån har någon inverkan på risken för läckage från tarmskarven. Därför kombinerades kolorektalcancerregistret med läkemedelsregistret så att alla patienter som erhöll en tarmskarv vid operation för ändtarmscancer och som stod på medicinering för diabetes, mot åderförkalkning eller nitroglycerin för hjärtat, under åren 2007 till 2010 bildade en kohort med förhöjd risk för hjärt-kärlsjukdom. Vi utförde sedan analyser där vi kontrollerade för ett antal bakgrundsfaktorer för att se om nivån på kärldelningen i sig påverkade risken för läckage från tarmskarven. Dessutom tittade vi separat på några undergrupper, inklusive de enskilda läkemedelsgrupperna separat och på patienter med manifest hjärt-kärlsjukdom, utifrån uppgifter i patientregistret, samt vid kombinationen manifest sjukdom och tecken till ökad samsjuklighet.

Delarbete II var en mer beskrivande studie, som undersökte hur vanligt det är med död efter tarmskarvande kirurgi för ändtarmscancer idag, hur vanligt det är med läckage från tarmskarven, och slutligen huruvida död efter kirurgi ökar vid läckage från skarven. Studieperioden var 2007–16 och död inom 90 dagar användes som utfall, eftersom bl.a. modern intensivvård möjliggör upprätthållande av vitala funktioner under längre tid än bara 30 dagar, den period som klassiskt har använts för att fånga död till följd av kirurgi. Dessutom gjordes en separat analys av de tarmskarvar som krävde reintervention – antingen operation eller röntgenledd punktion – till följd av läckage. All data

(13)

hämtades från kolorektalcancerregistret utom data angående död, som hämtades från totalbefolkningsregistret. Vi studerade även om olika faktorer kunde påverka huruvida läckage ökade risken för död i så kallade interaktionsanalyser.

I delarbete III återgick vi till att studera nivån på kärldelningen, men denna gång om den kunde påverka de långsiktiga resultat som själva kirurgin syftar till, såsom överlevad och återfall i ändtarmscancer, samt huruvida fler lymfkörtlar kunde avlägsnas med en mer central kärldelning. Alla patienter som i botande syfte genomgick operation under 2007–14 ingick. Data hämtades från kolorektalcancerregistret, totalbefolkningsregistret, dödsorsaksregistret och the Human Mortality Database. För att få två likartade grupper användes en statistisk metod som kallas propensity score matching. Den går ut på att para ihop varje patient som fick en hög kärldelning med en som fick en låg kärldelning, men där matchningen görs så att patienterna rent statistiskt har samma sannolikhet att få en hög kärldelning (trots att hälften alltså i praktiken fick en låg). Därmed är det bara den faktiska kärldelningsnivån som skiljer dem åt. Då kan man göra analyser på en matchad studiepopulation, med bara kärldelningsnivå som förklarande faktor, samt åskådliggöra skillnaderna gällande överlevnad och återfall över tid med så kallade Kaplan-Meierkurvor och utvärdera om de skiljer sig signifikant åt. Även här gjordes ett antal analyser av delar av studiepopulationen separat.

Det fjärde delarbetet designades, till skillnad från de övriga tre delarbetena, inte för att studera orsak och verkan utan för att se huruvida smärta efter operation kunde förutspå risken för läckage från tarmskarven efter tjock- eller ändtarmscancerkirurgi. Alla patienter som återfanns i både kolorektalcancerregistret och det svenska perioperativa registret år 2014–17 inkluderades, om de hade erhållit en tarmskarv i samband med operation. På uppvakningsavdelningen mäts normalt patientens smärta och registreras därmed i det perioperativa registret, där vi kunde uthämta den maximala smärtan som patienterna hade rapporterat. Sedan undersökte vi huruvida smärtan, som ursprungligen rapporteras på en skala från 0–10 men i huvudanalyserna grupperades som låg (0–3) eller hög (4–10), var kopplad till läckagefrekvensen. Dessutom indelades smärtan i 3 grupper (0–3, 4–7 och 8–10) för att se om kraftigare smärta gav högre risk för läckage. Dag 1 efter operation har man nämligen som kirurg – eller annan sjukvårdspersonal för den delen – inte så mycket att gå på eftersom blodprover ännu inte har hunnit börja stiga och patientens normala funktioner sällan är igång igen. Studien fokuserade därför på smärta, särskilt som det finns en stor potentiell vinst i att hitta patienter med läckage innan de blir kraftigt påverkade.

(14)

Resultat

Inklusion och exklusion av patienter i delstudie I ledde till en slutgiltig studiepopulation på 722 patienter, varav 334 opererades med hög kärldelning och 388 med låg, och av vilka totalt 82 patienter drabbades av läckage – 41 i respektive grupp. Det tycks överlag som om kärldelningsnivån inte påverkar risken för läckage och när vi i en basal och mer utförlig modell justerade för störfaktorer blev oddskvoten mellan hög och låg kärldelningsnivå allt närmre 1, vilket talar för samma risk i bägge grupper. Däremot verkar det finnas en ökad risk för skarvläckage hos patienter som opereras med hög kärldelning på ett sjukhus som inte utför lika mycket kirurgi. Likaså tycks den allra sjukaste delen av studiepopulationen – patienter med både manifest hjärt-kärlsjukdom och riklig samsjuklighet – drabbas av läckage i större omfattning efter hög än efter låg kärldelning. För både lågvolymssjukhus och den sjukaste undergruppen var oddskvoten runt 4, vilket innebär en påtagligt ökad risk, om fynden i studien överensstämmer med verkligheten.

Delstudie II kunde, bland de 6948 patienter som ingick i studiepopulationen, visa hur 693 patienter (10,0%) drabbades av läckage från tarmskarven. För dessa var dödligheten 3,9% efter operation medan den bara var 1,5% för de som inte drabbades av läckage. Men de 294 patienter (4,2%) som behövde åtgärdas med någon form av ingrepp p.g.a. läckage hade en dödlighet på 7,5%. Därmed visade studien hur läckage fortfarande är vanligt och associerat med ökad dödlighet, men effekten drivs enbart av de läckage som är så pass allvarliga att de kräver reintervention. När vi undersökte de patienter med läckage som inte behövde reintervention hade de faktiskt lägre dödlighet, om än statistiskt negligerbart, än patienter utan tecken till läckage. Dessutom visade sig ålder vara en tydlig riskfaktor för ökad dödlighet i samband med läckage.

Studiepopulationen i den tredje delstudien bestod av 8287 patienter, av vilka 4968 (60,0%) opererades med låg kärldelning och 3049 (36,8%) med hög. För de resterande 270 patienterna (3,3%) saknades data om kärldelningsnivå. Cancer- specifik 5-årsöverlevnad var 86% överlag och total 5-årsöverlevnad 76% på den matchade studiepopulationen, medan lokala och generella återfall inom samma tidsperiod var 2,6% respektive 18%. Ingen skillnad kunde skönjas beroende på om patienter var opererade med hög eller låg kärldelning. Däremot innebar hög kärldelning en diskret ökad total lymfkörtelskörd (17,7 vs 16,7 körtlar), medan antalet cancerdrabbade lymfkörtlar som avlägsnades var närmast identiskt. När vi gjorde om analyserna på den omatchade studiepopulationen, men kontrollerade för störfaktorer, fick vi likaledes negativa resultat.

Delstudie IV bestod slutligen av 3084 patienter, av vilka 189 (6,1%) drabbades av läckage och 121 (3,9%) behövde reopereras p.g.a. detta. Smärta på

(15)

uppvakningsavdelningen var kopplad till risk för läckage och reoperation, med oddskvot 1,73 respektive 2,13, när vi kontrollerade för störfaktorer. När smärtan indelades i tre grupper uppvisade gruppen med högst smärta (8–10) motsvarande oddskvoter på 2,42 respektive 2,61. Således tycks ökad smärta redan tidigt efter kirurgi kunna förbättra prediktionen av läckage och behovet av reoperation, och en biologisk gradient tycks finnas där patienter med högre smärtskattning hade högre risk för läckage.

Slutsatser

Läckage från tarmskarvar efter ändtarmskirurgi är även i modern sjukvård en vanligt förekommande komplikation som ökar risken för död efter kirurgi. Denna risk tycks dock vara mest påtaglig vid mer allvarligt läckage, som fordrar mer åtgärder än bara antibiotikabehandling. Risken att drabbas av läckage tycks däremot inte påverkas av valet av kärldelningsnivå. För de allra sjukaste patienterna är dock hög kärldelning eventuellt mer riskfylld och bör nog endast övervägas i särskilda fall, särskilt som vi inte kunde finna några fördelar med hög kärldelning gällande långtidsöverlevnad eller återfall i ändtarmscancer i studiepopulationen i sin helhet. Därmed borde patientens och kirurgens egna preferenser kunna styra valet av kärldelningsnivå. Eftersom läckage dock fortsätter att vara vanligt förekommande, och emellanåt innebär ödesdigra konsekvenser, är behovet av tidig upptäckt av patienter med läckage av vikt. Det tycks som om smärta kan förutspå risken för läckage och borde, tillsammans med andra avvikelser från den normala postoperative återhämtningen, kunna användas för att avgöra vilka patienter som behöver ytterligare utredning.

(16)

List of dissertation publications

Study I

High arterial ligation and risk of anastomotic leakage in anterior resection for rectal cancer in patients with increased cardiovascular risk. Petrus Boström, Markku M Haapamäki, Peter Matthiessen, Rickard Ljung, Jörgen Rutegård, Martin Rutegård Colorectal Disease. 2015 17(11):1018–27

Study II

Population-based cohort study of the impact on postoperative mortality of anastomotic leakage after anterior resection for rectal cancer. Petrus Boström, Markku M Haapamäki, Jörgen Rutegård, Peter Matthiessen, Martin Rutegård BJS Open. 2018 15;3(1):106–111

Study III

Oncological impact of high vascular tie after

surgery for rectal cancer: a nation-wide cohort study. Petrus Boström, Daniel Kverneng Hultberg, Jenny Häggström, Markku M. Haapamäki, Peter Matthiessen, Jörgen Rutegård, Martin Rutegård. Annals of Surgery. 2019.

Volume publish ahead of print

Study IV

Early postoperative pain as a predictor of anastomotic leakage in colorectal cancer surgery. Petrus Boström, Johan Svensson, Camilla Brorsson, Martin Rutegård (submitted)

(17)

Related publications

Arterial ligation in anterior resection for rectal cancer: A validation study of the Swedish Colorectal Cancer Registry. Petrus Boström, Jörgen Rutegård, Markku M Haapamäki, Peter Matthiessen, Martin Rutegård. Acta Oncol. 2014 Jul;53(7):892–7

Current use of diverting stoma in anterior resection for cancer: a population-based cohort study of total and partial mesorectal excision. Martin Rutegård, Petrus Boström, Markku M Haapamäki, Peter Matthiessen, Jörgen Rutegård. Int J Colorectal Dis. 2016 Mar;31(3):579–85

(18)

(19)

Background

A brief history of rectal cancer

In the beginning, there was indeed darkness, as colorectal cancer was a death sentence. Even worse, a rectal tumour’s close proximity to the pelvic bones and other richly innervated structures often resulted in great pain and morbidity during the process. Moreover, the first surgical attempts to treat the condition did not change this outcome, as local excisions with a transperineal approach, performed in the early 19^th century, still resulted in death in generalised carcinomatosis within a couple of years(1).

After 1845, when the efficacy of ether was famously demonstrated at Massachusetts General Hospital by William Morton(2), general anaesthesia rapidly became available, permitting more radical surgery, and the first abdominal operations under general anaesthesia were carried out soon afterwards. However, the lack of antibiotics and blood transfusions still limited the extent of the resection, which partly explains the extreme 3-year recurrence rates of 80–100% even among some very prominent rectal cancer surgeons, including Billroth, Kraske, Kocher and Miles, in the early 20^th century(1, 3). But failure is a strong driving force, and in 1908 Moynihan hypothesised the now well-established paradigm that sound oncological surgery must follow the lymphatic anatomy(4, 5). Simultaneously, and in accordance with this belief, the abdominoperineal excision of the rectum (APE), still in use today, was introduced by Miles(3), although Czerny had reported a similar operation as early as 1883, but with much greater recurrence rates(6).

Two scientific debates regarding rectal cancer research thus began. One is omnipresent in surgery and revolves around the distribution of emphasis on short-term harm and long-term benefit. At the beginning of the 20^th century, this was manifested in the early scepticism towards the combination of laparotomy and transperineal approach of APE, an operation that initially exhibited postoperative mortality in the range of 30–42% compared to 3–14% for contemporary perineal methods(3, 7). With time, however, the postoperative mortality rates decreased substantially, with Gabriel reporting 2.6% 30-day mortality, and the method became more widely accepted(8). Yet the balance between postoperative morbidity and mortality versus oncological results is as vividly discussed today as it has ever been, especially after the recent emergence of non-surgical treatment with curative intent, described by Habr-Gama and colleagues(9).

(20)

Figure 1. The rectum and surrounding tissue in the coronal plane, including a low rectal tumour. The dashed line marks the border of an APE.

The other old controversy of importance to this dissertation regards the decision on how centrally to ligate the feeding vessel. While Moynihan advocated ligation of the inferior mesenteric artery, flush to the aorta, in order to achieve a greater lymph node harvest(4, 5), Miles, on the contrary, seems to have supported ligation of the artery distal to the take-off of the left colic artery(3). After numerous studies and meta-analyses covering more than a century, the best option is yet to be determined(10, 11).

Initially, gastrointestinal continuity was rare after rectal cancer surgery, partly due to the importance of short operative times and the lack of treatment for the septic complications of anastomotic leakage(8). However, it was also in part due to Miles’ misperception that downward spread of the tumour was as important as upward spread, and the whole rectum and anal canal therefore needed to be removed(3, 4). This misunderstanding of the lymphatic spread gradually weakened and, in the middle of the 20^th century, Goligher noted that among 1,500 APE specimens, only 2% demonstrated evidence of downward spread, almost all of which also showed extensive upward spread(12). This led to the anterior resection and its colorectal anastomosis gaining traction, especially through the Mayo Clinic(13). The name anterior resection stems from the abdominal

(21)

approach, in contrast to posterior resection, which was the old transperineal surgery, with or without transection of the sacrum(14).

The last large paradigm shift in the surgical treatment of rectal cancer occurred after Heald in 1982 published his technique for performing rectal cancer surgery using total mesorectal excsision (TME), in which the whole mesorectum is excised en bloc with the rectal tumour. It was, once again, triggered by a new understanding of the relevant anatomy, in which the mesorectum was considered the main reason behind local recurrence(15). Previously, the surgeon’s hand had defined the outer border of the pelvic resection, using blunt dissection, usually not under direct vision, thus violating the modern principle of performing surgery in natural anatomical-embryological planes. With the advent of TME surgery, local recurrence rates were again greatly reduced, with Heald reporting a 4-year local recurrence rate of 3.7% in an era when 20% would be acceptable(16). The concept was implemented on a large scale in the Stockholm TME project, where education of rectal surgeons resulted in diminishing the 5-year recurrence rate from 14% to 6%(17).

Figure 2. The male patient’s rectum and surrounding organs in the sagittal plane. The dashed line marks the excision border in TME surgery.

In addition to the surgical evolution, modern medicine has also invented and developed medical therapies for cancer treatment. Parallel to the development of TME surgery, neoadjuvant radiotherapy became a mainstay of the multimodal treatment of rectal cancer, in part driven by Swedish randomised controlled trials (RCTs). The Stockholm I trial demonstrated a decrease in the local recurrence rate from 30% to 16% after a median of 107 months, if radiotherapy was given

(22)

before curative surgery. However, the postoperative mortality increased from 2%

to 8%, which could explain the lack of survival benefit demonstrated(18). The Stockholm II trial followed, which was a population-based study in which the local recurrence rates of 25% versus 12% again favoured irradiation before curative surgery(19). The Dutch TME trial corroborated these findings, showing a local recurrence reduction from 6% to 1% for stage II and 15% to 4% for stage III rectal cancer(20), while the German Rectal Cancer Study Group showed the efficacy of neoadjuvant chemoradiotherapy over adjuvant treatment, with lower local recurrence levels and increased rates of sphincter-sparing surgery(21).

And here we are today. The modern management of rectal cancer is built upon a vast array of therapeutic options, including endoscopy, irradiation, chemotherapy, antibody-based therapy, immune therapy, radiofrequency ablation, oncogenetic counselling and even watchful waiting. Nonetheless, it remains indeed a surgical disease, where most patients undergo surgery, which is also the most commonly used modality, especially if the intent is curative(22).

Surgery is, however, not without risk. Abdominal surgery for rectal cancer is a major operation, with non-trivial mortality and considerable morbidity(23). In addition, the availability of a larger treatment arsenal compared to a century ago, as well as the impact of other treatment modalities on the surgical outcome, add to the complexity of the treatment decisions. In this rapidly evolving field of surgery, intense research efforts are needed to help guide both surgeons and patients.

Definitions

The word rectum, meaning “straight” in Latin(24), describes the portion of the large intestine between the end of the sigmoid colon and the beginning of the anal canal. In vivo, however, the rectum by means of the mesorectum closely follows the sacral curvature and is also angled anteriorly by the puborectal muscle distally, therefore being straight only in the coronal view or as a surgical specimen(25). Its exact definition has evaded consensus, partly due to the different needs of and techniques available to different medical specialties and partly due to biological variation between individuals and the lack of a discrete anatomical transition. Not surprisingly, different definitions yield widely different rectal lengths. Today, cross-sectional pelvic magnetic resonance imaging is considered the most objective method of defining the rectum, due its high fidelity, availability, reproducibility and lack of operator-dependency(26).

Experts have mostly agreed on using the sigmoid take-off as the formal boundary between the sigmoid colon and the rectum(27). It is the point at which the broad and mobile sigmoid mesocolon becomes the narrower and more tethered mesorectum(25). This dissertation, however, follows the most commonly used pragmatic definition, in which the adult rectum is defined as the large bowel

(23)

within 15 cm from the anal verge, by rigid endoscopy(21, 26-29). And an adenocarcinoma of mucosal origin, the lower border of which is within these 15 cm, is considered a rectal cancer, following the Swedish standard(28).

Histology

Histologically, the rectum is similar to the colon, being comprised of a mucosa, submucosa and two muscular layers, with longitudinal fibres running outside an inner, circular layer. In addition, ganglia belonging to the enteric nervous system are present in the submucosa and between the muscular layers(30). The circular layer becomes more prominent at three distinct locations – the valves of Houston.

Of these sickle-shaped, mucosa-lined intrusions into the lumen, the middle valve is most prominent on endoscopy and marks the beginning of the ampulla recti, a widening of the rectal lumen. Distally, the columnar epithelium of the mucosa is replaced by transitional epithelium of the anal canal at the level of the pelvic diaphragm, making the limits of the rectum less arbitrary distally than proximally(25). In distinction to the colon, the rectum is only partly covered by peritoneum, which initially ends posteriorly, then on the sides of the rectum, and is finally reflected so that the third portion is completely extraperitoneal(31). In addition, the longitudinal muscular layer of the colon forms condensations, termed taeniae, which together with the appendices epiploicae disappear as the sigmoid colon becomes the rectum(27).

Vessels and nerves

The main arterial supply to the rectum originates from the inferior mesenteric artery (IMA), which at some point during its descent is given the name superior rectal artery (SRA), usually after the take-off of both the left colic artery (LCA) and the sigmoid arteries(32), even though in Sweden the name SRA is sometimes given to the vessel directly after the take-off of the LCA. The SRA initially runs dorsal to the rectum, before branching into two arteries running on either side of the rectum. In addition, vascular supply is derived from the internal iliac artery, both directly in the form of the middle rectal artery, though often diminutive or absent(33), and from the inferior rectal artery, which is a branch of the internal pudendal artery(34). The venous drainage mirrors the arteries close to the bowel, but the superior rectal vein drains into the inferior mesenteric vein and from there to the portal vein, instead of coursing towards the systemic veins(35). The lymphatic drainage of the rectum down to the level of the dentate line primarily follows the SRA towards the preaortic nodes, whereas drainage below the dentate line leads to superficial inguinal nodes. However, the distal third of the rectum seems to display more variation, with lymph drainage in both these directions as well as towards the internal iliac lymph nodes, by colorectal surgeons often termed pelvic lateral sidewall lymph nodes, which also seem to include the

(24)

obturator lymph nodes(36). These nodes are sometimes affected even in early rectal cancer and in Japan are termed N3 status and routinely removed if the rectal tumour invades through the muscularis propria and the lower border is below the peritoneal reflection, while they are still considered metastatic and not locoregional in Western practice(36-39).

Figure 3. Arteries, veins and lymphatic drainage of the rectum.

The rectum is surrounded by the mesorectum, a structure consisting mainly of adipose and connective tissue in which the superior rectal vessels and lymphatic drainage course. It is bulkier posteriorly where it has a groove in the posterior midline, yielding it a bi-lobar appearance, as it follows the median sacral crest and the concave shape of the levator muscles on either side of the crest. The mesorectum is in turn enveloped in a thin visceral fascia propria of the rectum, which fuses with the parietal endopelvic fascia anteriorly to form Denonvilliers’

fascia, more pronounced in men, separating the mesorectum from the urogenital organs. Waldeyer’s fascia is the name given to the posterior parietal endopelvic fascia, which distally fuses with the visceral fascia to form the rectosacral ligament distal to the fourth sacral body(31, 32, 37). The importance of the mesorectum was highlighted with the invention of TME surgery, as surgery in the

(25)

holy plane, between the visceral and parietal fascia, enables an oncologically sound operation with limited morbidity(15, 16).

The sympathetic nerve supply to the pelvic organs including the rectum originates at the T10 to L2 level, coursing via the sympathetic trunk, converging at the level of the take-off of the IMA and then forming the superior hypogastric plexus just below the aortic bifurcation(31, 40). From there it branches into the paired hypogastric nerves, which descend along the pelvic sidewalls, posterior to the presacral fascia, and are joined in the pelvis by the nervi erigentes, parasympathetic fibres of the sacral plexus, originally from S2 to S4, to form the inferior hypogastric plexuses(35, 41). In males, this plexus gives rise to the neurovascular bundle of Walsh, which travels just anterior to the holy plane on the posterolateral aspect of the prostate(37).

Figure 4. The important innervation in proximity to the rectum. Originally from Incision (incision.care). Re-printed with permission.

Damage to the sympathetic nerve fibres could result in urinary and faecal incontinence or disturbances, anejaculation or retrograde ejaculation, while damage to the parasympathetic nerves could lead to urinary retention and erectile dysfunction in both men and women, including vaginal dryness(42).

Anatomically, however, all important nerves run outside of the preferred surgical plane, hence invalidating Goligher’s famous words on the balance between oncological and functional outcome(43): “If you haven’t made the patient impotent, you probably haven’t cured him of rectal cancer.”

(26)

Physiology

While the colon’s function is to regulate electrolytes, absorb water and some nutrients, as well as to harbour the microflora needed to perform these tasks, the rectum mainly acts as a reservoir and emptying mechanism for faeces(30, 35).

This is accomplished by an intricate process requiring both sensory input and motor output, where the rectum, anus and pelvic floor act in a coordinated fashion, involving both locoregional reflexes and cerebral input. Increasing filling volume causes a distention of the rectal wall, the sensory input of which results in a desire to defecate and, when socially convenient, enables the conscious relaxation of the external anal sphincter and the puborectalis muscle, which reduces the anorectal angle. The rectum is thereby straightened and opened, allowing for rectal contractions to completely evacuate its contents(35, 44).

Afterwards, continence is again achieved, a delicate task which includes the voluntary elimination of intestinal gas while at the same time preserving continence for liquid stool, which is attributed to the highly sensitive mucosa above the dentate line, as well as a local reflex between the mucosa of the distal anal canal and the external sphincter muscle(35, 45).

Carcinogenesis

The majority of adenocarcinomas follow the adenoma-carcinoma sequence(46), where invasion of the submucosa marks the transformation to a carcinoma, which ends with the ability to metastasise(35, 46, 47). Though the concept of cancer as a genetic disease was familiar in the middle of the 20^th century, the seminal paper by Nowell in 1976 outlined the theory of clonal expansion of a single cell. This is a Darwinian process, where acquired genetic variability through mutation leads to sequential selection of increasingly aggressive sublines(48). Our understanding of the molecular basis for this increasing genomic instability increased with time, so that a number of common genetic and epigenetic changes and the order in which they tend to occur have been described, including the APC, KRAS, TP53, BRAF oncogene and DNA mismatch repair genes(35). This provides us with a greater understanding of the carcinogenesis, increases our predictive ability on an individual level and allows for novel and individualised treatments(49, 50). One of the most important aspects of the adenoma-carcinoma sequence is the fact that cancer precursors and early stages could be detected by radiology or endoscopy, since the development towards an invasive colorectal carcinoma often represents a decade of clonal expansion(49).

However, as adenomas are found in approximately 40% of a Western population undergoing endoscopy, but only 3% suffer from adenocarcinoma, meaning that most precursors do not develop into a carcinoma(46).

(27)

Epidemiology

Measuring by age-standardised incidence and mortality rates worldwide, colorectal cancer is the fourth most common cancer type, after breast, prostate and lung cancer, and the third most common cause of cancer death, surpassed only by breast and lung cancer(51). It is also in many regards a disease of the industrialised world, with a 10-fold difference between countries with the highest to the lowest incidences(28). In Sweden, colorectal cancer has surpassed lung cancer in incidence and breast cancer in mortality and today represents every 9^th patient who is diagnosed with a malignancy(52). Rectal cancer constitutes approximately one third of all colorectal cancer, with a seemingly stable age- standardised incidence rate over time, recently with approximately 2000 new Swedish cases per year. It has become a disease with a stable or slightly decreasing mortality depending on which time period is used for comparison, since the greater reduction in mortality occurred in the 1970’s to the middle of the 1990’s. This results in a 66% 5-year overall survival, and a 56% corresponding relative survival(22, 53, 54). As with many cancer types, the strongest risk factor is increasing age, which is the reason for the need for age-standardised epidemiological measures(22, 55).

General risk factors

Unlike colon cancer, which is evenly distributed between men and women, rectal cancer has a male preponderance. In Sweden, the yearly age-standardised incidence is 25 per 100,000 men and 15 per 100,000 women, and the corresponding mortality rate 10 and 5 per 100,000(22). The reason for this sex disparity is not completely known, in part because risk factors were historically evaluated for colorectal cancer overall, not stratified by tumour site(56). Since colorectal cancer is more common in the industrialised world, it has been hypothesised that a diet which relies relatively more on animal fat and red meat than fibres and vegetables would be more carcinogenic(57). The original theory concerned faecal transit times, since fibres decrease the time in which the gastrointestinal canal is exposed to possible faecal carcinogens, and Burkitt noted that the Bantu people produced very bulky stools and displayed a low incidence of colorectal cancer(58). However, their diet is not the only thing that separates the Bantu people of South Africa from a Western population and studies since then have both corroborated and negated different dietary causes of colorectal cancer(59-61). The metabolic status, including diabetes mellitus, obesity and physical activity, has also been thoroughly examined and though there seems to be a link, it is generally weaker for rectal than colon cancer(62-66). In contrast, smoking seems to more strongly associated with rectal than colon cancer(67, 68).

(28)

Specific risk factors

In contrast to the weak associations between life style risk factors and rectal cancer, inflammatory bowel disease and some hereditary conditions display strong links to rectal cancer(69, 70). Especially ulcerative colitis is associated with rectal cancer and the severity of the inflammation correlates with increased risk, while the association between Crohn’s disease and rectal cancer is more modest.

Both diseases, however, show a stronger link to rectal than colon cancer(71-73).

The Lynch syndrome, originally described by Warthin in 1913(74) and later rediscovered by Lynch in 1966(75) is an autosomal dominant non-polyposis hereditary syndrome. It is associated with a number of malignancies but mostly with colorectal cancer, being the most common cause of hereditary colorectal cancer. A number of genetic mutations behind the disease have been identified, most of which occur in mismatch repair genes(76). The other hereditary syndrome associated with colorectal cancer is familial adenomatous polyposis, in which affected families have a large number polyps and ensuing increased risk of colorectal cancer(77), due to mutations in the APC gene(47). It is also an autosomal dominant disease associated with extracolonic manifestations(78).

Overall, germline mutations are presumably present in 5–10% of all cases of colorectal cancer, though this estimate probably will increase as our molecular understanding of the disease deepens(79, 80).

Modern management of rectal cancer

Staging

The diagnosis of colorectal cancer is made either by screening endoscopy, incidentally by radiology, or initiated due to symptoms of the disease. The most predictive symptom is rectal bleeding, especially if combined with changes in bowel habits or abdominal pain(81). Colon cancer often present with more vague symptoms compared to rectal cancer, partly due to the fact that bleeding is usually occult(82). After diagnosis, the work-up aims to provide a correct tumour staging in order to yield a prognosis and devise a treatment plan.

In 1932, Dukes presented the first formal histological system for staging colorectal cancer, designed to increase objectivity and prognostic ability, especially for patients undergoing surgery(83, 84). The system has been modified over the years and is still in use today, but often replaced by the TNM Classification of malignant tumour system, introduced by the American Joint Committee on Cancer and the International Union Against Cancer, and nowadays endorsed by the World Health Organization(85, 86).

(29)

TNM, now in its 8^th edition, receives its acronym from tumour, (lymph) nodes and (distant) metastases, respectively, containing a clinical (cTNM), pathological (pTNM), and post-neoadjuvant (ypTNM) version(87). The clinical staging is more arbitrary, using whichever diagnostic tools are available, often including radiology and biopsy. Pathological staging adds a standardised in-depth histopathological analysis of the surgical specimen, as well as intraoperative findings(88). In practice, cTNM is used to guide the choice of surgery (including no surgery) and neoadjuvant therapy, while pTNM is used for predicting long- term outcomes and selecting adjuvant therapy. In Sweden, the TNM system is used, complemented by WHO’s classification of tumours of the digestive system, currently in its 5^th edition(28, 87, 89), yielding the following tables:

Table 1. Stages of colorectal cancer. A. Primary tumour stages. B. Regional lymph node stages. C. Distant metastasis stages. D. Tumour stage.²

A. Primary tumour (T) TX Non-assessible

T0 No signs of a primary tumour

Tis Carcinoma in situ: intraepithelial or invasion into lamina propria T1 Tumour invasion into the submucosa

Subclassification of T1

Sm1 Invasion into 1/3 of the submucosa Sm2 Invasion into 2/3 of the submucosa Sm3 Invasion into 3/3 of the submucosa T2 Tumour invades muscularis propria

T3 Tumour invades into pericolorectal tissue Subclassification of T3

T3a <1 mm invasion outside of muscularis propria T3b 1–5 mm invasion outside of muscularis propria T3c 5–15 mm invasion outside of muscularis propria T3d > 15 mm invasion outside of muscularis propria T4a Tumour penetrates the surface of visceral peritoneum T4b Tumour invades other organs or structures

(30)

B. Regional lymph nodes (N)

Nx Nodal status unknown or non-assessible N0 No regional lymph node metastases N1 Metastases in 1–3 regional lymph nodes N1a Metastasis in 1 regional lymph nodes N1b Metastases in 2–3 regional lymph nodes

N1c Tumour deposits in subserosa, mesenterium or non-peritonealised pericolic or perirectal tissue, without evidence of regional lymph node metastases

N2 Metastases in >3 regional lymph nodes N2a Metastases in 4–6 regional lymph nodes N2b Metastases in >6 regional lymph nodes C. Distant metastases (M)

M0 No distant metastases M1 Any distant metastasis

M1a Metastasis to an organ or tissue without peritoneal metastases M1b Metastases to >1 organ/tissue

M1c Metastases to peritoneum, with or without other metastases D. Tumour stage (TNM)

Stage T N M

0 Tis N0 M0

I T1–T2 N0 M0

II T3–T4 N0 M0

III Any T N1–2 M0

IV Any T Any N M1

In addition to the TNM staging, the standardised histopathological report on the surgical specimen should include information on mucinous or conventional adenocarcinoma, completeness of the mesorectal envelope, classification of surgical radicality (including circumferential resection margin), regression grade, tumour deposits, and perineural, lymphatic or vascular invasion (including extramural vascular invasion). Moreover, molecular and immunohistochemical pathology analyses are being increasingly utilised, to further increase the predictive abilities and thereby govern the choice of adjuvant therapy(28).

(31)

Figure 5. Magnetic resonance imaging of a rectal tumour with positive lymph nodes and peritoneal metastases. The curved shape of the rectum along the sacrum is evident in the sagittal plane.³

Oncological treatment

Accumulated clinical experience and scientific discoveries have led to an individualised multimodality management of colorectal cancer in general and rectal cancer in particular(50). As discussed earlier, radiotherapy has become a cornerstone in the treatment of rectal cancer, while it has failed to show any advantage even in locally advanced colon cancer(35, 90). Strong evidence also supports its use in a neoadjuvant instead of an adjuvant setting(21). However, while the influential German Rectal Cancer Group Study by Sauer proved the efficacy of long course radiotherapy, given as 28 fractions of 1.8 Gray (Gy) to a total of 50.4 Gy, with surgery after 6–8 weeks, the original Swedish trials used short course radiotherapy, fractioned as 5x5 Gy, followed by surgery within one week, the latter not only showing a major decrease in local recurrence but also an overall 5-year survival advantage, 58% versus 48%(18, 21, 91). In addition, the Stockholm III trial showed no difference regarding oncological outcomes or postoperative complications between a long course and a pooled group of short

3Originally from Radiology assistant – Rectal cancer MR staging by van Loenhout, Rhiannon, et al.

Creative commons.

(32)

course radiotherapy protocols(92). This has led to different praxis in different countries, where the Nordic countries tend to favour short course radiotherapy.

According to the present Swedish guidelines, the only indication for long course radiotherapy is as part of chemoradiotherapy for technically challenging T4b tumours, while the indication for short course therapy grows stronger with especially increasing tumour stage and proximity to the anal verge(28). One disadvantage with short course radiotherapy is that there is insufficient time for downsizing and downstaging, which in Sweden led to the introduction of a longer waiting period, with re-evaluation after 6–8 weeks, before attempting surgery(92).

For locally advanced tumours, long course radiotherapy is often utilised, and the addition of neoadjuvant chemotherapy seems to increase the locoregional control, as evidenced by both lower locoregional recurrence rates and a higher proportion of sphincter-saving surgery(21, 93, 94). For colon cancer, neoadjuvant chemotherapy is becoming increasingly popular in locally advanced disease, sometimes even given as chemoradiotherapy(95-97). However, the most common indication for chemotherapy is in the adjuvant setting, where colon and rectal cancer is more uniformly treated, and studies have showed that the benefits outweigh the harm in stage III cancer or stage II cancer with certain additional risk factors(35). In addition, chemotherapy and immune therapy are used to treat metastatic colorectal cancer in a palliative setting(28).

Watch and wait

Another great treatment shift is presently occurring in rectal cancer, however, with the introduction of the watch and wait strategy from Brazil(9). Some rectal cancers respond so well to radiotherapy, at times combined with local excision, that it challenges the century-old paradigm of ontogenetic surgery of the lymphatic anatomy. When complete clinical response has been achieved with local treatment, more and more patients have simply been monitored closely for recurrence, without an obvious need for abdominal surgery. More long-term data is rendered(98, 99), increasing our understanding of tumour biology, hopefully allowing for an individualised selection of patients in need of abdominal surgery.

One issue with the watch and wait protocol is that it contradicts the present Swedish attempt to lower the proportion of patients given radiotherapy, since the only patients for whom we have data on the benefits of a conservative approach are stage I and II tumours(9), the few patients for whom radiotherapy often would be obviated(28).

Local and endoscopic excisions

The surgical treatment dogma has in a way come full circle with the increasing use of local excision techniques for colorectal cancer, though today patients are

(33)

carefully selected by radiology, histology, endoscopy and patient characteristics and preferences. Avoiding the morbidity and mortality risk of an abdominal surgical resection is especially advantageous in the old and frail. The first described successful transanal submucosal excision was performed by Parks in 1966(35), but it was only after the introduction of the transanal endoscopic microsurgery (TEM) system in 1979 that the whole rectum could be visualized and reached, which led to an increased use and larger proportions of negative resection margins(100-102). The flexible platform, transanal minimally invasive surgery, which was introduced in 2009, is an evolution of this technique, used for polyps, T1 and some T2 tumours, but data on long-term outcomes are still largely missing(103).

Colonoscopic techniques for local excision have also improved rapidly in recent years, used mainly for colonic but sometimes including rectal tumours. While polyps are usually removed with simple polypectomy, endoscopic mucosal resection (EMR) is a technique in which the submucosa is infiltrated, allowing even flat lesion to be raised and removed by a snare, preferably en bloc if there is suspicion of cancer. In Japan, the more demanding technique of endoscopic submucosal dissection (ESD) was originally developed for gastric lesions but commonly used for colorectal lesions, including cancer in situ or T1a tumours(104). The lesion is raised by infiltrating the submucosa, after which a colonoscopic knife with coagulative properties is used to remove the lesion en bloc. EMR is easier, faster, cheaper and has a lower risk of perforation, while ESD instead has the benefit of a significantly lower risk of bleeding and most importantly local recurrence, a major driving force for interval cancers(105, 106).

These partial wall resections were in 2015 accompanied by endoscopic full thickness resection, using over-the-scope clips to close the gap, though this technique is mainly used for re-resection of adenomas, where scarring from earlier resection tethers the mucosa to the underlying layers of the wall(35, 107).

Abdominal surgery

As mentioned earlier, surgery is still the main curative treatment of colorectal cancer(35), being both the most efficient and cost-effective treatment(108, 109).

For colon cancer, this often results in a right or left hemicolectomy or a sigmoid resection, where laparoscopy during the last decade has replaced open surgery as the default access strategy, though locally advanced tumours requiring multivisceral resection and some transverse colon tumours are still usually treated by open surgery(35, 108, 110). In addition, following the same ontogenetic and oncological principles as TME for rectal cancer, German colon cancer surgeons have developed the concept of complete mesocolic excision (CME) including central vascular ligation(111), similar to the Japanese D3 dissection for colon cancer(112). The promising results from the case series in Erlangen, where

(34)

a drop from 6.5% to 3.6% 5-year locoregional recurrence rate was seen(111), was initially not uniformly reproduced elsewhere, but seems to be gaining evidence, though the increased morbidity of CME surgery could potentially offset any oncological benefits(113-116).

Figure 6. High ligation of the IMA flush to the aorta is seen during an open left hemicolectomy. The small bowel is retracted towards the patient’s right side, out of view.⁴

For rectal cancer, there are three major surgical procedures performed today, namely anterior resection, abdominoperineal resection and Hartmanns’s procedure. In Sweden, anterior resection is still the most common procedure, performed for roughly 30% of all rectal cancers in Sweden, with abdominoperineal resection comprising 25% and Hartmann’s procedure 10%.

TEM or local excision is performed in 2.5% of cases, while 30% are left without surgical treatment(22). The steady trend over the last decades is a slow decline in the proportion of anterior resections performed, and the increasing non-surgical

4 Originally from Total mesorectal excision and sphincter preservation - the early steps of rectal cancer surgery, J. Coloproctol. (Rio J.) vol.34 no.1 2014, by Perez, Rodrigo, et al Creative commons.

(35)

treatment during the same period, with other surgical modalities more or less constant(22). The main indication for anterior resection is a cancer in the upper or middle rectum. If the tumour is located so that a 5 cm distal margin could be achieved, then a partial mesorectal excision could be performed, with an end-to- end anastomosis, and probably without the use of a diverting stoma, though this is still added in approximately one third of partial mesorectal excision operations performed in Sweden(117-119). For more distal tumours, TME surgery is performed as described earlier, removing the rectum and mesorectum en bloc, with a distal margin sometimes of as little as 1 cm, and the ensuing anastomosis performed as either a side-to-end anastomosis or a reservoir. These anastomoses are regularly protected by a defunctioning stoma, as frequently as 92% in Swedish data(119).

For tumours in the distal rectum, APE is most commonly performed. Indications for APE could be both oncological and functional, the latter including both a poor sphincter function or the presence of comorbidities which increase the risk and sequelae of anastomotic leakage. When performed for oncological reasons, the procedure is not seldom performed as the more extensive extralevator APE, as advocated by Torbjörn Holm and colleagues in Stockholm, in which the perineal dissection includes the levator muscle on one or both sides(120). This development was partly a result of the inferior outcomes regarding recurrence and survival after traditional APE, when compared to anterior resection(20, 121, 122). The waist of the surgical specimen at the level of the tumour was suspected to be responsible for this observation, and the extralevator APE was in part invented to correct this inadequate dissection margin. However, the morbidity of the surgery is also greater, since the ensuing defect is larger and none of the replacement alternatives – mainly a biological mesh or a muscle flap – is perfect(123, 124). These morbidity issues and recent data, suggesting an increase in positive circumferential resection margins with extralevator compared to conventional APE, highlight the need for further research(125, 126).

The alternative to APE, if performed for non-oncological reasons, is Hartmann’s procedure, where the distal rectum is stapled and left in situ, while an end- sigmoidostomy is performed after removal of the tumour. The operative trauma is thereby smaller than for APE. However, even though no anastomosis is performed, both pelvic sepsis and abscesses have been observed as often as in 12–

19% of cases(127, 128), presumably due to frailty and old age being common in this patient group(28). A Swedish ongoing multicentre RCT on Hartmann’s procedure versus intersphincteric APE is still recruiting patients, trying to establish the relative risks of each procedure(129). Meanwhile, a well-designed, prospective, observational, multicentre study from England recently demonstrated similar overall complication rates but increased medical sequelae and reduced quality of life after APE(130).

(36)

Anastomotic leakage

The main disadvantage of anterior resection for rectal cancer is the risk of anastomotic leakage, which causes both mortality and great morbidity(131-133), including impaired functional outcome(134, 135), and possibly recurrence and decreased survival(136-138).

Figure 7. The construction of a bowel anastomoses, as it would have been constructed in the early 20^th century.⁵

The definition of anastomotic leakage has traditionally varied greatly in both scientific publications and general practice(139). The International Study Group of Rectal Cancer in 2010 therefore proposed a standardised definition and grading of anastomotic leakage after rectal cancer surgery. They recommended to define as leakage any communication between the intra- and extraluminal compartments owing to a defect of the integrity of the intestinal wall at the anastomosis, including the suture or staple line of a neorectal reservoir, or a pelvic abscess(140). In addition, they suggested a grading system dependent on the clinical consequences, in which a grade A leakage is not associated with any therapeutic measure, while a grade B leakage is associated with any intervention except for laparotomy, which marks the transition to a grade C leakage. The idea is to catch the phenomenon of what was earlier described as radiological leakage in grade A, indicating a leakage that does not affect the patient’s vital signs,

5 Originally from American practice of surgery; a complete system of the science and art of surgery, by Bryant, Joseph D (1906), through digitising sponsor Open Knowledge Commons. Creative commons

(37)

laboratory values or subjective well-being, and therefore is considered of no clinical importance, and simultaneously distinguish smaller, symptomatic leakages from postoperative disasters. In Sweden, even though the Swedish Colorectal Cancer Registry (SCRCR) does not explicitly mention the definition of leakage, common practice is to include only symptomatic leakages(141). The ambiguity of the definition is, however, problematic to all research based on these data, including this dissertation.

Figure 8. Anastomotic leakage after anterior resection for rectal cancer with an irritated rectal wall in proximity to the pus coming out through the defect in the anastomosis.

The manner in which anastomotic leakage is diagnosed hence partly relates to its grade. Today, with the widespread use of computer tomography, most leakages are diagnosed radiologically(142, 143), which enables detection of all grades of leakage. However, radiology is usually not routinely scheduled, and sometimes falsely negative(143), so vigilance is warranted when clinical signs or laboratory values do not follow the expected path of an uneventful postoperative period.

Tachycardia and fever predates the diagnosis in more than half of patients who eventually suffer from leakage, though the full spectrum of deranged vital signs which accompanies sepsis could be seen, as well as less obvious signs such as a purulent anal discharge or functional urinary tract symptoms(142). Increased inflammatory markers are seen, including procalcitonin and C-reactive protein, where the greatest area under the curve for absolute values for seems to be

(38)

around postoperative day 4(144-146), whereas the recently published PREDICT study indicates a better prognostic ability if the patient’s own trajectory is used for reference(147). In addition, a number of proinflammatory markers have been evaluated in drain fluid but the decreasing use of drains limit their usability in general practice(148, 149). The lack of known early warning signs prompted this dissertation’s investigation into the association of postoperative pain and leakage.

Risk factors for anastomotic leakage

The risk of anastomotic leakage is not uniformly distributed along the gastrointestinal tract. In general, anastomoses including the beginning or end of the tract yield leakage rates which are multiple times higher than when two intraperitoneal organs are anastomosed. Hence, approximately 1 in 10 anastomoses involving the oeshopagus or rectum tend to leak, but only 1 in 50 involving the stomach or small bowel. The only gastrointestinal operations with higher leak rates are pancreatic procedures, where the degradative contents of bile and pancreatic fluid presumably contribute to leakage being seen after 1 in every 4 procedures(142, 150, 151). Due to the high incidence and possible catastrophic sequelae of anastomotic leakage after rectal cancer surgery, much research activity of the past has focused on risk factors and protective measures, in order to prevent or at least mitigate and treat its effects. With the introduction of neoadjuvant radiotherapy, the incidence of anastomotic leakage seems to have increased, though the association is not always corroborated(152, 153), while the evidence of its increase after the introduction of TME surgery is less ambiguous(154-156). Hypothetically, the tissue oedema, pelvic fibrosis and necrosis of irradiation could negatively impact the healing properties of the anastomosis. This is one potent argument in favour of adding radiotherapy for resectable rectal cancer only in the adjuvant setting, as has been more common in American practice(157). TME surgery, on the other hand, could in theory result in decreased perfusion to the healing anastomosis, since the mesorectum with its blood vessels is fully removed. However, though modifiable, both risk factors have been shown to be oncologically advantageous(18, 20, 21). Hence, one important aim of future research is to better delineate the border between overall harm and benefit, paving the way to a more selective use of these treatment modalities.

Male sex and obesity are similar, well established, risk factors for leakage after rectal cancer surgery(158-160). Both factors increase the difficulty of the surgery performed, where the narrow and deep male pelvis is often clinically considered more technically problematic than large amounts of intraabdominal fat.

However, both factors are also independently associated with leakage after non- pelvic surgery, such as colon cancer(161), which could be an indication of a less anatomic mechanism of effect. For rectal cancer, however, the difficult dissection

Rectal Cancer

Rectal Cancer

The Influence of Surgical Technique on Morbidity, Mortality and Survival

Petrus Boström

Table of Contents

Abstract

Abbreviations

Populärvetenskaplig sammanfattning

List of dissertation publications

Related publications

Background