Maria Antonietta Gambacorta - Congressi · PDF fileAttualità nei trattamenti integrati...

AttualitaneitrattamentiintegratidelcarcinomadelRettolocalmenteavanzato:versolapreservazioned’organoIlpuntodivistadelradioterapistaoncologo

MariaAntoniettaGambacorta

APPROPRIATEZZADELL’IMAGING

NEITUMORIDELRETTO

pCR andOrgan Preservation

• Themeaning ofCR

• Howtoachieve CR• What toirradiate• Howtopredict

pCR after CRT

Protocol pCR Concomitant CT CumulativepCR

EORTC22921 13.7*

5FUFFCD9203 11.4

STAR 16

5FU

and

oxaliplatin

≈15%ACCORD 19.2

NASBP R04 20.9

CAO-ARO-AIO 04 16.5

*pT0

LC-CRT

The meaning of CR after CRT

pCR andlongterm outcomes:RCT

Articles

986 www.thelancet.com/oncology Vol 16 August 2015

445 patients who actually received adjuvant fl uorouracil and leucovorin and oxaliplatin, as compared with 100 (21%) of 470 patients who actually received adjuvant fl uorouracil alone. Grade 3 or worse diarrhoea was the most frequent chronic adverse eff ect in both groups (31 [7%] in the investigational group vs 42 [9%] in the

control group). The incidence of grade 3–4 sensory neuropathy in the fl uorouracil and leucovorin and oxaliplatin group decreased from 41 (10%) patients during treatment to 13 patients (3%) at 1 year follow-up. Among the 326 patients in the fl uorouracil and leucovorin and oxaliplatin group having sphincter-sparing surgery,

Figure 4: Disease-free survival in the intention-to-treat population by patient subgroups according to pretreatment and surgical or pathological factors after preoperative chemoradiotherapyThe size of the quadrats represents the proportion of patients. HR=hazard ratio. ECOG=Eastern Cooperative Oncology Group.

52/23058/23849/145

112/43447/179

115/48343/123

61/249

83/302

14/55

145/57114/41

33/146123/452

12/1149/39

29/16092/260

9/17

75/416

45/133

30/42

135/567

6/15

9/10419/14840/15457/154

98/398

5/3145/151

159/613

84/24164/23350/149

149/44049/183

136/47558/141

87/216

84/336

24/64

169/56927/50

58/159134/451

6/837/39

41/183120/27817/26

94/423

53/131

44/60

167/584

7/9

6/8130/17648/14871/169

105/416

10/3067/152

198/623

HR (95% CI)

0·61 (0·43–0·86)0·87 (0·61–1·24)1·06 (0·71–1·58)

0·73 (0·57–0·93)0·98 (0·65–1·46)

0·80 (0·62–1·02)0·86 (0·58–1·28)

0·57 (0·41–0·79)

1·10 (0·81–1·49)

0·64 (0·33–1·25)

0·83 (0·67–1·04)0·62 (0·32–1·18)

0·56 (0·36–0·86)0·91 (0·71–1·16)

1·52 (0·57–4·05)

1·40 (0·52–3·76)

0·77 (0·48–1·24)

0·78 (0·60–1·03)

0·76 (0·34–1·70)

0·78 (0·58–1·06)

0·82 (0·55–1·22)

1·09 (0·65–1·81)

0·80 (0·64–1·01)

0·40 (0·13–1·25)

1·20 (0·43–3·36)0·72 (0·40–1·28)0·74 (0·49–1·13)0·89 (0·63–1·28)

0·98 (0·74–1·29)

0·46 (0·16–1·35)0·58 (0·39–0·85)

0·79 (0·64–0·98)

0·2 0·5 1·0 2·0 3·0 4·0Hazard ratio

Age (years)<6161–70>70

SexMaleFemale

ECOG performance status01–2

Location from anal verge0–5 cm>5–10 cm>10 cm

cT categorycT2–3cT4

cN categorycN0cN+

ypT categoryypT0ypTis/T1ypT2ypT3ypT4

ypN categoryypN0ypN1ypN2

Completeness of local tumour resectionR0R1

TNM stageypT0N0Stage IStage IIStage III

Type of surgeryLow anterior resectionIntersphincteric resectionAbdominoperineal resection

TotalAll

Favours investigational group Favours control group

Investigational group(events/n)

Control group(events/n)

Rodel C et al. Lancet Oncol 2015

pCR andlongterm outcomes:

pooled analyses

Maas M et al. Lancet Oncol 2010

cT3orcT2low

Long Course RT-CT

Surgery descalation

Pucciarellietal,S.etalDis ColonRectum 2013GérardJPetal.JClin Oncol 2014Vuong Tetal.Semin ColonRectal Surg 2010Maas MetalJClin oncol 2011Appelt A.LancetOncol 2015

cCR

Sexual function (male*)

Stephens etalJCO2010

surgery

RT

*Only 11%ofwomen completed thequestionnaire at 2yrs

Sphincter function

Unintentional releaseofstool

Stephens etalJCO2010


• Themeaning ofCR• Howtoachieve CR• What toirradiate• Howtopredict

Howtoachieve CR

• Chemotherapy intensification

• Radiotherapy intensification

• TIMECRT Surgery

CTintensification

During RT

‘New’CTagents

Targettherapies

Sequence oftherapies

TotalNeoadjuvant Therapy

CTintensification during RT:

‘Newagents’

BytheCourtesyofC.Rodel – ESTROCourse2016

Targettherapies

By the Courtesy of C.Rodel – ESTRO Course 2016

Targettherapies

CTintensification:

TotalNeoadjuvant Treatment

Fernandez-Martoz Cetal.JCO2010

Conclusion: Both treatment approaches yield similar outcomes. Given the lower acute toxicity and improved compli-ance with induction CT compared with adjuvant CT, integrating effective systemic therapy before CRT and surgery is apromising strategy and should be examined in phase III trials.Key words: locally advanced rectal cancer, induction chemotherapy, phase II randomized trial, adjuvant chemotherapy

introductionThe role of adjuvant chemotherapy (CT) after preoperative che-moradiation (CRT) has not been well established in publishedrandomized trials [1–4]. The low compliance with CT followingpreoperative CRT and surgery as well as the lack of a reliablemethod to select patients, might contribute to these negativeresults. Furthermore, the British Chronicle and the DutchSCRIPT-PROCTOR trials were terminated early because pooraccrual and consequent insufficient statistical power.More recently, the Korean ADORE multicenter, randomized,

phase II study included poor prognostic rectal cancer patientswho received preoperative CRT and had pathological stage II(ypT3-4N0) or III (ypTanyN1-2) disease. Three-year disease-free survival (DFS) was significantly improved in patientstreated with adjuvant FOLFOX compared with those receivingadjuvant 5-fluorouracil (5-FU) and leucovorin (71.6% versus62.9%, P = 0.047). Interestingly, 96% of patients completed theplanned number of adjuvant CT cycles [5].

An alternative approach to improving outcome is to deliver in-duction CT before preoperative CRT. Induction CT may be asso-ciated with better treatment compliance and may allow fullsystemic doses of CT to be delivered. Our group completed a ran-domized phase II trial (GCR-3) comparing this approach withconventional preoperative CRT followed by surgery and post-operative adjuvant CT. Primary results were reported in 2010.Compared with postoperative adjuvant capecitabine oxaliplatin(CAPOX), induction CAPOX before CRT had similar pathologic-al complete response (pCR) and complete resection rates while atthe same time achieving more favorable compliance and toxicityprofiles [6]. Given these advantages, this induction strategy isnow considered a new treatment alternative for patients withhigh-risk rectal cancer in some European institutions and theUnited States [7]. We report long-term outcomes including localrecurrence, distant recurrence and survival results after a medianfollow-up of 69 months.

108 patients randomized

49 (94%) patients commenced CRT

3 ineligible patients excluded

46 (89%) patients underwent surgery

45 R0 resections

1 R1 resection

37 (71%) patients commenced adjuvant CT

9 did not receive any cycle

7 received 1-2 cycles

2 received 3 cycles

28 received 4 cycles

54 (96%) patients commenced CRT

54 (96%) patients commenced induction CT

1 received 1 cycle

1 received 3 cycles

52 received 4 cycles

54 (96%) patients underwent surgery

48 R0 resections

2 R1 resections

2 R 2 resections

2 ineligible patients excluded

Follow-up07/2013

Life Status Know n = 55Tumor status know = 55

Follow-up07/2013

Life Status Know n = 50Tumor status know = 50

Arm B: Induction CT56 pts 100%

CT CTRT Surgery

Arm A: Adjuvant CT52 pts 100%

CTRT Surgery CT

Figure 1. Consort diagram. CRT, chemoradiation; CT, chemotherapy.

Volume 26 | No. 8 | August 2015 doi:10.1093/annonc/mdv223 |

Annals of Oncology original articles

by guest on Decem

ber 10, 2016http://annonc.oxfordjournals.org/

Dow

nloaded from

CT)—are listed in Table 1. The predominant clinical disease stage wascT3N!. After random assignment, arm B included more patientswith T4 lesions than arm A (13% v 6%) and also more grade 3 tumors(11% v 2%). Arm A included more patients with a threatened circum-ferential resection margin than arm B (10% v 0%). Otherwise, patientcharacteristics were well balanced between the two arms.

The number of eligible patients who commenced treatment was49 (94%) of 52 in arm A and 54 (96%) of 56 in arm B. Overall, 46patients (89%) in arm A and 54 (96%) in arm B underwent surgery.When patients who had the appropriate dose reductions were in-cluded, more patients in arm B than arm A completed the study as perprotocol (91% v 54%; P " .0001). A total of nine patients (17%) inarm A and 1 patient (2%) in arm B discontinued study treatmentbecause of adverse events (P # .006). The percentages of patients whodiscontinued treatment for other reasons were similar in the twogroups (Table 2).

Efficacy ParametersA total of 46 patients in arm A, after preoperative CRT, and 54

patients in arm B, after induction CT and preoperative CRT, under-went surgery. On the basis of an intent-to-treat analysis, a pCR(ypT0N0M0) was achieved in seven patients in arm A (13%; CI 95%,5.6% to 25.8%) and in eight patients in arm B (14%; CI 95%, 6.4% to

26.2%). Downstaging (defined as lower pathologic T stage comparedwith the pretreatment clinical T stage) was observed in 30 patients(58%) in arm A and in 24 patients (43%) in arm B. An R0 resectionwas achieved in 45 patients (87%) in arm A and in 48 patients (86%) inarm B. In arm A, the tumor regression grade (TRG) of the primarytumor (ypT0) was TRG 4 in seven patients, and an additional 22patients showed tumor regression of greater than 50% of the tumormass (TRG 3). In contrast, in arm B, eight patients had a TRG of 4, and20 patients had a TRG of 3 (Table 3).

Table 1. Baseline Demographic and Clinical Characteristics for theTotal Patient Group

Characteristic

Arm A:Post-

operativeAdjuvant

CT(n # 52)

Arm B:Induction

CT(n # 56)

No. % No. %

Age, yearsMedian 62 60Range 42-75 38-76

SexMale 34 65 39 70Female 18 35 17 30

ECOG status0 36 69 33 591 15 29 22 392 — 1 2Unknown 1 2 —

Locally advanced rectal cancer definitionby MRI $ US

cT4 resectable 3 6 7 13cT3 lower third (! 6 cm from anal

verge) tumors 12 23 18 32CRM threatened or involved,

mid-rectal cancer 5 10 —Any cT3N! 31 59 31 55Missing 1 2 —

Pathologic gradeNot otherwise specified 12 23 11 201: well differentiated 12 23 11 202: moderately differentiated 27 52 28 503: poorly differentiated 1 2 6 11

Abbreviations: CT, chemotherapy; ECOG, Eastern Cooperative OncologyGroup; MRI, magnetic resonance imagine; US, ultrasound; CRM, circumfer-ential resection margin.

Table 2. Treatment Received and Reasons for Discontinuation

Variable

Arm A:Post-

operativeAdjuvant

CT(n # 52)

Arm B:Induction

CT(n # 56)

PNo. % No. %

Completion of study treatment perprotocol 28 54 51 91 " .0001

Discontinuation of study treatment! 22 44 5 9 " .001Reason for discontinuation

Progression of disease 1 2 0 .93Adverse event 9 17 1 2 .006Investigator decision/unfit for CT 4 6 0 .05Death 3 6 3 6 1Toxicity 2 4 2 4Other† 1 2 1 2Consent withdrawn 5 10 1 2 .1

Abbreviation: CT, chemotherapy.!Two patients in arm A did not have data available.†One patient in arm A committed suicide; one patient in arm B had

major depression.

Table 3. End Points for the Total Patient Group

End Point

Arm A:Post-

operativeAdjuvant

CT(n # 52)

Arm B:Induction

CT(n # 56)

P !No. % No. %

pCR 7 13 8 14 .9495% CI, % 5.6 to 25.8 6.4 to 26.2

Downstaging 30 58 24 43 .1395% CI, % 43.2 to 71.3 29.7 to 56.8

R0 resection rates 45 87 48 86 .40TRG†

4: complete regression 7 15 8 15 .883: % 50% of tumor mass 22 48 20 372: " 25%-50% of tumor mass 11 24 13 241: " 25% of tumor mass 2 4 3 60: no regression 1 2 3 6Not otherwise specified 3 7 7 13

Abbreviations: CT, chemotherapy; pCR, pathologic complete response; TRG,tumor regression grade.

!Fisher’s exact test.†The denominators were the patients who actually underwent resection.

Fernandez-Martos et al

862 © 2010 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 185.11.153.252 on December 10, 2016 from 185.011.153.252Copyright © 2016 American Society of Clinical Oncology. All rights reserved.

Preoperative treatmentintensification:


Fernandez-Martoz Cetal.JCO2010

Howtoachieve CR



• TIMECRT Surgery

RTintensification

D50,TRG1= 92.0Gy

D50,TRG1-2= 72.1Gy

Appelt ALetal.Int.JRadiat Oncol Biol Phys 2013

Reliability of RT for cure

Burbach etal.Radiother Oncol 2014

RT intensification: dose> 60Gy

Fig. 2. Meta-analysis forest plot of pCR-rates and pooled estimate in comparison to a reference line of control group (14.8%) [3] (pCR = pathological complete response).

6 Impact of radiotherapy boost on pathological complete response in rectal cancer

pCR 20%

since in some patients surgery was omitted for other reasons suchas a worsened condition, newly diagnosed metastasis or patient’srefusal. Fourth, the pathologic assessment was different betweenstudies, and therefore prone to bias. Ten of 14 included studiesstandardized assessment, of which 3 explicitly used Mandard’sscore [35]. Others only mentioned that one pathologist assessedif there was ‘absence of viable tumor cells’ in the specimen. Fifth,destruction of solitary tumor cells may continue long after termi-nation of radiotherapy, indicating that timing of surgery impactsresponse assessment. Three studies have shown increased pCR-rates when surgery was postponed from 8 to 11 weeks post-radia-tion (from 11.5 to 14.0%) [61], and when shorter surgical intervalsare compared to intervals of >6–8, or >7, weeks (from 13.7 to19.5%, and 16 to 28.0% respectively) [62,63]. A relative risk of1.42 (1.19–1.68) for pCR was reported for intervals longer than6–8 weeks as compared to intervals shorter than 6–8 weeks. Nev-ertheless, in our data we did not see an association between inter-val-length and pCR-rate, presumably because pCR-rate variedlargely at each interval length with only a few studies available

per interval-length point in the analysis. Such variation is common,and therefore often observed in systematic reviews on pCR-ratesfollowing CRT [1,3,64]. To further investigate the impact of pro-longed intervals on pCR-rate and sphincter preservation, severalrandomized clinical trials are currently recruiting (GRECCAR6/NCT01648894 [65] and NCT01037049). Nonetheless, if such pre-sumed time-effects allow extrapolation to when doses are esca-lated, pCR-rates and organ-preservation might even furtherbenefit when longer intervals prove to be safe. Sixth, only a singlestudy reported interval between radiotherapy and brachytherapy,which did not allow further meta-analysis. Finally, acceleratedtreatment (higher dose per fraction, i.e. simultaneous integratedboost) increases the biological effective dose which may benefitresponse [66,67], especially when tumor-regrowth time is short[68,69]. Nevertheless, some of these accelerated schedules remainchallenging because of considerable toxicity [19,24,70–72] andperi/post-operative complications [72,73]. It is likely that such tox-icity originates from irradiation of surrounding tissues instead ofthe tumor, as a result of a previously acquired treatment plan

Fig. 3. Forest plot of available acute grade P3 toxicity and resectability with pooled estimate.

J.P.M. Burbach et al. / Radiotherapy and Oncology 113 (2014) 1–9 7

G3-4TOX10%

RTintensification

Valentini V et al. ESTRO-33-2014

RTintensification

Modified from Valentini V et al. ESTRO-33-2014

pCR 24.4 23.8 ns

RTintensification

Valentini V et al. ESTRO-33-2014

Howtoachieve CR



• TIMECRT Surgery

Articles

www.thelancet.com/oncology Vol 16 August 2015 959

ProceduresThe trial consisted of a series of four sequential phase 2 study groups. In group 1, patients were treated with fl uorouracil-based chemoradiation and total mesorectal excision to establish the proportion of patients achieving a pathological complete response at baseline. In groups 2–4, patients received two, four, or six cycles of mFOLFOX6 between chemoradiation and total meso-rectal excision (fi gure 1). Patients in all study groups were treated with fl uorouracil 225 mg/m² per day by continuous infusion, 7 days per week throughout radiation. Fluorouracil infusion was given for 5−6 weeks, depending on the number of radiation boosts given. Radiation treatment was given once a day at 1·8 Gy per day, 5 days per week for 5 weeks, for a total of 45 Gy in 25 fractions, followed by a minimum boost of 5·4 Gy. In patients in whom the entire small bowel could be excluded from the fi nal cone down, a second boost of 3·6 Gy (54 Gy total cumulative dose) was given. A linear accelerator using a minimum 6 MV energy in three to four fi elds was delivered. Intensity-modulated radiation treatment was permitted if approved by the supervising radiation oncologist.

Patients in group 1 had surgery 6–8 weeks after chemo-radiation. Patients in groups 2–4 received cycles of mFOLFOX6 4−5 weeks after the completion of chemoradiation: patients in group 2 received two cycles, those in group 3 four cycles, and those in group 4 six cycles of mFOLFOX6. Each cycle consisted of racemic leucovorin 200 mg/m² or 400 mg/m², according to the discretion of the treating investigator, oxaliplatin

85 mg/m² in a 2-h infusion, bolus fl uorouracil 400 mg/m² on day 1, and a 46-h infusion of fl uorouracil 2400 mg/m². Patients had surgery 3–5 weeks after the last cycle of mFOLFOX6. To ensure patients in groups 2–4 were not placed at risk of disease progression during the lengthened chemoradiation-to-surgery interval, tumour response was assessed by Response Evaluation Criteria in Solid Tumors guidelines16 during the neoadjuvant treatment course. Patients with progressive or stable disease at interim assessment did not receive additional mFOLFOX6 and had total mesorectal excision without delay.

Surgery was done according to the principles of sharp mesorectal excision. Specimens were assessed according to the recommendations of the Association of Directors of Anatomic and Surgical Pathology.17 Postoperative chemotherapy to complete a total of eight cycles of mFOLFOX6 was recommended, but not dictated by the trial, and was delivered at the discretion of the treating physician.

OutcomesThe primary endpoint of the study was the proportion of patients achieving a pathological complete response (defi ned as the absence of tumour cells in the surgical specimen, both at the primary tumour site and at regional lymph nodes) in each study group. We also collected information on the proportion of patients who achieved pathological partial response (defi ned as having at least a 30% decrease in tumour width, or length in circumferential tumours), stable disease (between a 30% decrease and a 20% increase), and progressive disease (at

Figure 1: Trial protocolRadiotherapy was given 5 days per week for 5 weeks (arrows) for a total of 45 Gy with a minimum boost of 5·4 Gy. Fluorouracil was given as a 225 mg/m² per day continuous infusion for 7 days per week during radiation therapy for 5−6 weeks, depending on the number of radiation boosts given. mFOLFOX6 was given in 2-week cycles of leucovorin 200 mg/m² or 400 mg/m² and oxaliplatin 85 mg/m² in a 2-h infusion, bolus fl uorouracil 400 mg/m² on day 1, and a 46-h infusion of fl uorouracil 2400 mg/m². *Interim assessments were done by proctoscopic examination; total mesorectal excision was done if the patient had stable or progressive disease.

Group 1

Group 2

Group 3

Group 4 Total mesorectal excision

Total mesorectal excision

Total mesorectal excision

Total mesorectal excisionContinuous infusion fluorouracil + radiotherapy

Continuous infusion fluorouracil + radiotherapy



Rest

Rest

Rest

Rest

Rest

Rest

Rest

mFOLFOX6 (four cycles)

mFOLFOX6 (two cycles)

mFOLFOX6 (six cycles)

0 2 4 6 8 10 12 14 16 18 20 22 24 26Weeks

*

*

*



Garcia-Aguilar J etal.LancetOncol 2015

Primary end-point:ypCR

RCà 12cmfromAVcT3-4,N0;cT,N1-2



Garcia-Aguilar J etal.LancetOncol 2015

RCà 12cmfromAVcT3-4,N0;cT,N1-2

259enrolled patients

Articles


during chemoradiation did not diff er between study groups (data not shown).

The proportion of patients experiencing adverse events during mFOLFOX6 treatment increased from group 2 to group 4. In group 2, two (3%) of 67 patients had grade 3 adverse events and one (1%) had a grade 4 adverse event; in group 3, 12 (18%) of 67 patients had grade 3 adverse events; in group 4, 18 (28%) of 65 patients had grade 3 adverse events and fi ve (8%) had grade 4 adverse events. The most common grade 3 or higher adverse events from mFOLFOX6 across study groups 2−4 were neutropenia in 11 patients (6%; fi ve in group 3 and six in group 4) and lymphopenia in seven patients (4%; three in group 3 and four in group 4). 18 (9%) of patients experienced neuropathy during mFOLFOX6 treatment, all of which were either grade 1 or 2. One patient had grade 1 neuropathy in group 2, six had grade 1 and one had grade 2 neuropathy in group 3, and nine had grade 1 and one had grade 2 neuropathy in group 4.

Table 2 summarises surgical results. The proportion of patients who received a sphincter-saving surgery and resection with negative margins was not signifi cantly diff erent between study groups (p=0·68 and p=0·089, respectively). The number of nodes examined and estimated blood loss were similar across all study groups (p=0·20 and p=0·62, respectively). Pelvic fi brosis, as measured by surgeon scoring from 1 (none) to 10 (maximum), increased in groups 2–4 (p=0·0001). However, the technical diffi culty of the operation, as

scored by the surgeon, was not signifi cantly diff erent across study groups (p=0·80).

No patient died during or after surgery in any study group. There was no signifi cant diff erence in the number of grade 3 or worse complications across study groups (all p>0·1; table 4). Grade 3 or worse complications were noted for nine (15%) patients in group 1, four (6%) patients in group 2, three (4%) patients in group 3, and six (9%) patients in group 4, when counting the maximum Clavien-Dindo grade complication for each patient. Of the 25 grade 3 or worse complications reported across all study groups, the most common were pelvic abscesses (seven patients: three in group 1, two in group 3, and two in group 4) and anastomotic leaks (seven patients: three in group 1, one in group 2, one in group 3, and two in group 4).

Using univariable logistic regression, we assessed known clinically relevant variables and study groups and their association with pathological complete response. The comparison of group 4 (the most intense regimen) with group 1 (the standard neoadjuvant regimen) showed a signifi cant association with pathological complete response (p=0·028; table 5). We then used multivariable logistic regression to model the probability of pathological complete response, examining whether a treatment eff ect on pathological complete response was present after adjusting for other known clinically relevant variables. In an intention-to-treat analysis that included study group, radiation dose, tumour stage, size, and distance to anal verge as variables, we found study group to be the only signifi cant predictor of pathological complete response (p=0·048; table 5). Patients in group 4 were signifi cantly more likely to achieve a pathological complete response than were patients in group 1 (odds ratio 3·49, 95% CI 1·39–8·75; p=0·011). In preplanned analyses, we tested two additional models in which we substituted study group for the treatment delivered as measured by cycles of mFOLFOX6 or the chemo radiation-to-surgery interval; both were signifi cant predictors of pathological complete response (p=0·028 and p=0·018, respectively).

Group 1 (n=60) Group 2 (n=67) Group 3 (n=67) Group 4 (n=65) p value

Time from start of chemoradiation to surgery (weeks) 14·2 (4·3) 17·1 (2·9) 21·0 (2·7) 25·2 (4·0) 0·0001

Time from end of chemoradiation to surgery (weeks) 8·5 (4·2) 11·1 (2·9) 15·4 (2·6) 19·3 (4·2) 0·0001

Sphincter-saving surgery 46 (77%) 50 (75%) 50 (75%) 44 (68%) 0·68

Ileostomy 38/46 (83%) 43/50 (86%) 47/50 (94%) 38/43 (88%)* 0·33

Resection with negative margins 59 (98%) 67 (100%) 64 (96%) 64 (100%)† 0·089

Number of nodes examined 12 (2–31) 14 (2–30) 13 (2–30) 11 (1–47) 0·20

Pelvic fi brosis‡ 2·4 (1·7) 3·9 (2·6) 4·4 (2·4) 3·9 (2·4) 0·0001

Technical diffi culty§ 4·6 (2·7) 4·9 (2·8) 5·1 (2·5) 4·8 (2·4) 0·80

Estimated blood loss (mL) 200 (50–1200) 225 (25–1500) 200 (50–1000) 150 (0–1000) 0·62

Data are mean (SD), number (%), n/N (%), or median (range). p values test the null hypothesis of equal means or proportions across study groups. *Information on whether an ileostomy was created or not was not available for one patient. †Data missing for one patient. ‡Scale ranges from 1 (none) to 10 (maximum). §Scale ranges from 1 (easy) to 10 (diffi cult).

Table 2: Surgical results

Group 1 (n=60)

Group 2 (n=67)

Group 3 (n=67)

Group 4 (n=65)

p value

Pathological complete response 11 (18%) 17 (25%) 20 (30%) 25 (38%) 0·0036

Partial response 44 (73%) 50 (75%) 46 (69%) 39 (60%) ··

Stable disease 5 (8%) 0 1 (1%) 1 (2%) ··

Data are number (%). p value tests the null hypothesis of equal proportions across study groups.

Table 3: Pathological tum our response

Primary end-point:ypCR

Articles


during chemoradiation did not diff er between study groups (data not shown).

The proportion of patients experiencing adverse events during mFOLFOX6 treatment increased from group 2 to group 4. In group 2, two (3%) of 67 patients had grade 3 adverse events and one (1%) had a grade 4 adverse event; in group 3, 12 (18%) of 67 patients had grade 3 adverse events; in group 4, 18 (28%) of 65 patients had grade 3 adverse events and fi ve (8%) had grade 4 adverse events. The most common grade 3 or higher adverse events from mFOLFOX6 across study groups 2−4 were neutropenia in 11 patients (6%; fi ve in group 3 and six in group 4) and lymphopenia in seven patients (4%; three in group 3 and four in group 4). 18 (9%) of patients experienced neuropathy during mFOLFOX6 treatment, all of which were either grade 1 or 2. One patient had grade 1 neuropathy in group 2, six had grade 1 and one had grade 2 neuropathy in group 3, and nine had grade 1 and one had grade 2 neuropathy in group 4.

Table 2 summarises surgical results. The proportion of patients who received a sphincter-saving surgery and resection with negative margins was not signifi cantly diff erent between study groups (p=0·68 and p=0·089, respectively). The number of nodes examined and estimated blood loss were similar across all study groups (p=0·20 and p=0·62, respectively). Pelvic fi brosis, as measured by surgeon scoring from 1 (none) to 10 (maximum), increased in groups 2–4 (p=0·0001). However, the technical diffi culty of the operation, as

scored by the surgeon, was not signifi cantly diff erent across study groups (p=0·80).

No patient died during or after surgery in any study group. There was no signifi cant diff erence in the number of grade 3 or worse complications across study groups (all p>0·1; table 4). Grade 3 or worse complications were noted for nine (15%) patients in group 1, four (6%) patients in group 2, three (4%) patients in group 3, and six (9%) patients in group 4, when counting the maximum Clavien-Dindo grade complication for each patient. Of the 25 grade 3 or worse complications reported across all study groups, the most common were pelvic abscesses (seven patients: three in group 1, two in group 3, and two in group 4) and anastomotic leaks (seven patients: three in group 1, one in group 2, one in group 3, and two in group 4).

Using univariable logistic regression, we assessed known clinically relevant variables and study groups and their association with pathological complete response. The comparison of group 4 (the most intense regimen) with group 1 (the standard neoadjuvant regimen) showed a signifi cant association with pathological complete response (p=0·028; table 5). We then used multivariable logistic regression to model the probability of pathological complete response, examining whether a treatment eff ect on pathological complete response was present after adjusting for other known clinically relevant variables. In an intention-to-treat analysis that included study group, radiation dose, tumour stage, size, and distance to anal verge as variables, we found study group to be the only signifi cant predictor of pathological complete response (p=0·048; table 5). Patients in group 4 were signifi cantly more likely to achieve a pathological complete response than were patients in group 1 (odds ratio 3·49, 95% CI 1·39–8·75; p=0·011). In preplanned analyses, we tested two additional models in which we substituted study group for the treatment delivered as measured by cycles of mFOLFOX6 or the chemo radiation-to-surgery interval; both were signifi cant predictors of pathological complete response (p=0·028 and p=0·018, respectively).

Group 1 (n=60) Group 2 (n=67) Group 3 (n=67) Group 4 (n=65) p value

Time from start of chemoradiation to surgery (weeks) 14·2 (4·3) 17·1 (2·9) 21·0 (2·7) 25·2 (4·0) 0·0001

Time from end of chemoradiation to surgery (weeks) 8·5 (4·2) 11·1 (2·9) 15·4 (2·6) 19·3 (4·2) 0·0001

Sphincter-saving surgery 46 (77%) 50 (75%) 50 (75%) 44 (68%) 0·68

Ileostomy 38/46 (83%) 43/50 (86%) 47/50 (94%) 38/43 (88%)* 0·33

Resection with negative margins 59 (98%) 67 (100%) 64 (96%) 64 (100%)† 0·089

Number of nodes examined 12 (2–31) 14 (2–30) 13 (2–30) 11 (1–47) 0·20

Pelvic fi brosis‡ 2·4 (1·7) 3·9 (2·6) 4·4 (2·4) 3·9 (2·4) 0·0001

Technical diffi culty§ 4·6 (2·7) 4·9 (2·8) 5·1 (2·5) 4·8 (2·4) 0·80

Estimated blood loss (mL) 200 (50–1200) 225 (25–1500) 200 (50–1000) 150 (0–1000) 0·62

Data are mean (SD), number (%), n/N (%), or median (range). p values test the null hypothesis of equal means or proportions across study groups. *Information on whether an ileostomy was created or not was not available for one patient. †Data missing for one patient. ‡Scale ranges from 1 (none) to 10 (maximum). §Scale ranges from 1 (easy) to 10 (diffi cult).

Table 2: Surgical results

Group 1 (n=60)

Group 2 (n=67)

Group 3 (n=67)

Group 4 (n=65)

p value

Pathological complete response 11 (18%) 17 (25%) 20 (30%) 25 (38%) 0·0036

Partial response 44 (73%) 50 (75%) 46 (69%) 39 (60%) ··

Stable disease 5 (8%) 0 1 (1%) 1 (2%) ··

Data are number (%). p value tests the null hypothesis of equal proportions across study groups.

Table 3: Pathological tum our response

Timeeffect?PooledanalysisofRCT:Accord12/0405

EORTC22921

FFCD9203

CAO/ARO/AIO-94

CAO-ARO-AIO-04

INTERACTandTROG01.04

5247ptsà 3078à 440(14%)hadpCR

MedianSurgeryTimeà 6wks

Shorter Interval

Group< 6wks (1953pts)

Longer Interval

Group>6wks (1125pts)

p

pCR 11% 19% .01

ValentiniVECCO2017

Plateau @ 16 months

95% @ 11 months

Timeeffect?

ValentiniVECCO2017

Timeeffect?

• Retrospective pooled analysis 2094pts from21Italian Centers

• Median timetosurgeryà 9weeks• CumulativepCRà 22.3%

< 6wks 7-12wks > 13wks p

pCR

AIRO

12.6% 23.0% 31.1% <0.001

pCRGarcia- Aguilar

18%(8.1wks) 25%(11wks) 30%(15wks) 0.036

MacchiaGetal.onbehalf ofAIRO-GIsubmitted


• Themeaning ofCR• Howtoachieve CR• What toirradiate

• Howtopredict

GTV + marginGTV + corresp mesorectum

Myerson et al IJROBP 2009

What toirradiate

3DConformal RT IMRT-IGRT

Imaging modality forGTV

(Pearson 0.24, p > 0.27) (Fig. 5). On T2, per ml volume increase theCI increased by 0.001 (SE 0.000, R2 0.367, p < 0.002) with interceptat 0.61. For Combi, per ml volume increase the CI increased by0.002 (SE 0.000, R2 0.371, p < 0.002) with intercept at 0.61.

Mean dCOMs per modality are presented in Fig. 6. T2 showed tohave the largest mean dCOM (3.30 mm) and Combi the lowest(2.57 mm), indicating T2 had most shape and location variability.Mean dCOMs were not significantly different between modalities(p > 0.70). dCOM variance was 2 times larger on T2 than on DWIor Combi, although one outlier existed on DWI. In retrospect,observers agreed the least experienced observer had delineatedan artifact which results in an unrealistically high dCOM value.

Contours were separated the widest on at least one point con-sidering the largest mean HD on T2 (18.59 mm, max 31.41 mm,min 9.16 mm) compared to DWI (mean HD 14.42 mm, min4.03 mm, max 49.10 mm) (Fig. 6). DWI again showed the outliercase (49.10 mm). On Combi, mean HD was between T2 and DWIvalues (Fig. 6). Repeated measures ANOVA showed a significantdifference in HD between modalities (p < 0.006), with a significantmean HD difference of 4.80 mm between T2 and DWI (p < 0.04) buta non-significant difference between T2 and Combi (p > 0.06) orDWI and Combi (p > 0.61). Separation over the whole contour-border was also largest on T2 (MD = 1.80 mm) and smallest onDWI (MD = 1.49 mm). Considering our voxel size ofP2.0 !P 2.0 mm on DWI, an average MD smaller than one voxelexists between observers. Again, the outlier case showed an

increased MD on DWI (9.05 mm). There were no significant differ-ences between modalities in MD (p > 0.19).

Discussion

In this study, two MRI sequences and their combination (T2 andDWI) were used for GTV-delineation in locally advanced rectal can-cer by three observers with the purpose of identifying GTV-delineation consistency and separation between contours. This isthe first study to actually calculate distance parameters, providingcrucial information for development of MRI-guided GTV booststrategies complementary to standard chemoradiation which isdirected to the entire CTV. Agreement between GTV delineationsperformed by the different observers was good and was compara-ble between different MRI sequences. Our results are in concor-dance with results of a previous rectal cancer delineation study[16]. Namely, the best consistency was reached on DWI images[16]. New findings of our study are inter-observer agreementimprovement with larger volumes, and comparable agreementafter combined T2-DWI sequences compared to both sequencesalone. This last finding may be due to the fact that some observerstend to rely mostly on T2-images while others prefer to trust theDWI-images more. Combining both scans, however, can alsointroduce a new inconsistency-factor since images have to be co-registered and topography can be different between sequences.

Fig. 1. Representative example of delineations in one patient of three observers on T2w (A and C) and DWI (B and D) sequences in transverse (A and B) and sagittal (C and D)direction.

402 Inter-observer agreement of MRI-based tumor delineation for preoperative radiotherapy boost in locally advanced rectal cancer

24pts

3observer

GTVon:• T2w

• DWI

• Combi T2-DWI

Bonnetain F etal.Eur J Cancer 2016

Imaging modality forGTV

of radical surgery. When watchful waiting is applied, the same MRsequences may not only play a role in delineation of tumors, butmay also be used to select patients with a cCR. For this purposeMR sequences are being optimized and new sequences are beingdeveloped, such as functional (DWI, MR spectroscopy) and veryhigh-resolution 7Tesla. These modalities are currently also underinvestigation in our institution [17,25] in order to provide a com-prehensive (MR-based) watchful waiting strategy to patients inthe future.

MRI-based delineation in rectal cancer is relatively new.Nowadays there is only one study available that investigatedinter-observer agreement [16]. Unfortunately, this study did not

calculate CIs or Fleiss’ kappa’s as measures of inter-observer agree-ment, but only compared volumes between observers (as dis-cussed above). In comparison with prostate cancer studies ondelineation agreement, we showed a considerably higher confor-mity on multiparametric (0.61) [14], T2w (0.61) and DWI (0.51)[15] imaging. This encourages to use MR for GTV delineationinstead of CT, which is the current standard for CTV delineation.Delineation of GTV volumes on CT resulted in contradictory vol-umes when compared to MRI GTV volumes [11,13], which is mostlikely dependent on worse soft-tissue contrast on CT. Whereas CTis primarily used to define CTVs, CT-based CTV delineationshowed comparable conformity indexes to what we obtained in

A B C0

50

100

150

200

T2A B

C D

Observers

Volu

me

[ml]

A B C0

50

100

150

DWI

Observers

Volu

me

[ml]

A B C0

50

100

150

Combi

Observers

Volu

me

[ml]

T2 DWI Combi0

50

100

150

200

All volumes per modality

Modality

Volu

me

[ml]

N Mean SD Min Max

T2 72 63.16 33.39 9.38 153.40DWI 69 41.94 25.49 7.60 115.18Combi 69 54.63 29.54 10.12 131.68

Fig. 2. Volumes per modality per observer (A–C), and for all observers combined per modality (mean, SD) (D).

0 50 100 150-20

0

20

40

60

80

A BAbsolute T2 and DWI volumes

Average volume [ml]

Mea

ndi

ffere

nce

[ml]

0 50 100 1500

2

4

Ratio between DWI to T2 volumes

Average volume [ml]

Diffe

renc

era

tio

1

3

Fig. 3. Bland–Altman plots representing absolute (A) and relative volume differences (B) between contours delineated by three observers on T2 and DWI.

404 Inter-observer agreement of MRI-based tumor delineation for preoperative radiotherapy boost in locally advanced rectal cancer

this study for MR-based GTV delineation [26]. Although theCT-based study by Nijkamp et al. was not aimed to identifydelineation-uncertainties for boost purposes, their presented

uncertainty-level indicates to what extent uncertainty is acceptedin current routine clinical practice. Nevertheless, there are somestudies that boost entire CTVs based on these uncertainties. They

Fleiss' kappa per modality

Flei

ss'k

appa

T2 DWI Combi0.5

0.6

0.7

0.8

0.9

1.0

N Mean SD Min Max N Mean SD Min Max

T2 24 .699 .078 .55 .84 T2 24 .820 .054 .71 .91DWI 23 .707 .104 .38 .85 DWI 23 .824 .079 .55 .92Combi 23 .690 .074 .53 .81 Combi 23 .815 .052 .69 .90

Conformity index per modality

Conf

orm

ityIn

dex

T2 DWI Combi0.0

0.2

0.4

0.6

0.8

1.0

A B

Fig. 4. Conformity index (A) and Fleiss’ kappa (B) values per modality for three observers.

T2

Volume [ml]

Conf

orm

ityIn

dex

0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0

DWI

Volume [ml]

Conf

orm

ityIn

dex

0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0

Combi

Volume [ml]

Conf

orm

ityIn

dex

0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0

Pearson p Reg. coeff. SE R2

0.606 <0.002 0.001 0.00 0.367


0.240 <0.27 - - -


0.609 <0.002 0.001 0.00 0.371

A B C

Fig. 5. Correlation between volumes and conformity indexes (IC’s) for T2 (A), DWI (B), and Combi (C) modalities.

dCOM per modality

Dist

ance

[mm

]

T2DWI

Combi-5

0

5

10

15

20

T2 DWI Combi0

20

40

60

HDmax per modality

Max

imal

dist

ance

[mm

]

T2 DWI Combi-5

0

5

10

Mean distance per modality

Mea

ndi

stan

ce[m

m]

Distance parameters (mm)

dCOM N Mean SD Min Max

T2 24 3.30 1.67 .79 6.83DWI 23 2.88 3.51 .63 17.74Combi 23 2.57 1.60 .77 5.39


MD N Mean SD Min Max

T2 24 1.61 .57 .75 2.79DWI 23 1.26 1.75 .29 9.05Combi 23 1.48 .56 .73 2.66


HDmax N Mean SD Min Max

T2 24 18.62 7.01 9.16 31.41DWI 23 13.82 9.06 4.03 49.10Combi 23 15.38 6.26 8.01 30.27



T2 24 18.59 7.00 9.16 31.41DWI 23 14.42 9.15 4.03 49.10Combi 23 15.42 6.24 8.01 30.27



T2 24 1.80 .60 .75 2.79DWI 23 1.49 1.80 .29 9.05Combi 23 1.68 .60 .73 2.66

A B C

Fig. 6. The distance between centers of mass (dCOM) (A), maximal Hausdorff distances (HDmax) (B), and mean distances (MD) (C) of contours within each observer-pair permodality.


this study for MR-based GTV delineation [26]. Although theCT-based study by Nijkamp et al. was not aimed to identifydelineation-uncertainties for boost purposes, their presented

uncertainty-level indicates to what extent uncertainty is acceptedin current routine clinical practice. Nevertheless, there are somestudies that boost entire CTVs based on these uncertainties. They

Fleiss' kappa per modality

Flei

ss'k

appa

T2 DWI Combi0.5

0.6

0.7

0.8

0.9

1.0

N Mean SD Min Max N Mean SD Min Max

T2 24 .699 .078 .55 .84 T2 24 .820 .054 .71 .91DWI 23 .707 .104 .38 .85 DWI 23 .824 .079 .55 .92Combi 23 .690 .074 .53 .81 Combi 23 .815 .052 .69 .90

Conformity index per modality

Conf

orm

ityIn

dex

T2 DWI Combi0.0

0.2

0.4

0.6

0.8

1.0

A B

Fig. 4. Conformity index (A) and Fleiss’ kappa (B) values per modality for three observers.

T2

Volume [ml]

Conf

orm

ityIn

dex

0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0

DWI

Volume [ml]

Conf

orm

ityIn

dex

0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0

Combi

Volume [ml]

Conf

orm

ityIn

dex

0 50 100 1500.0

0.2

0.4

0.6

0.8

1.0


0.606 <0.002 0.001 0.00 0.367


0.240 <0.27 - - -


0.609 <0.002 0.001 0.00 0.371

A B C

Fig. 5. Correlation between volumes and conformity indexes (IC’s) for T2 (A), DWI (B), and Combi (C) modalities.

dCOM per modality

Dist

ance

[mm

]

T2DWI

Combi-5

0

5

10

15

20

T2 DWI Combi0

20

40

60

HDmax per modality

Max

imal

dist

ance

[mm

]

T2 DWI Combi-5

0

5

10

Mean distance per modality

Mea

ndi

stan

ce[m

m]


dCOM N Mean SD Min Max

T2 24 3.30 1.67 .79 6.83DWI 23 2.88 3.51 .63 17.74Combi 23 2.57 1.60 .77 5.39



T2 24 1.61 .57 .75 2.79DWI 23 1.26 1.75 .29 9.05Combi 23 1.48 .56 .73 2.66



T2 24 18.62 7.01 9.16 31.41DWI 23 13.82 9.06 4.03 49.10Combi 23 15.38 6.26 8.01 30.27



T2 24 18.59 7.00 9.16 31.41DWI 23 14.42 9.15 4.03 49.10Combi 23 15.42 6.24 8.01 30.27



T2 24 1.80 .60 .75 2.79DWI 23 1.49 1.80 .29 9.05Combi 23 1.68 .60 .73 2.66

A B C

Fig. 6. The distance between centers of mass (dCOM) (A), maximal Hausdorff distances (HDmax) (B), and mean distances (MD) (C) of contours within each observer-pair permodality.


volume overlapping margins

Bonnetain F etal.Eur J Cancer 2016

Target motion

Shape

variation Set-up

•Prone up to 0.24 cm

•left-right direction

•Supine < 0.1 cm

Nijkamp J et al. Radiother Oncol 2009

AFTERTREATMENTBEFORE

IMAGE GUIDED RADIOTHERAPY

MRI HYBRID Machines


• Themeaning ofCR• Howtoachieve CR• What toirradiate• Howtopredict

CTPET

Nomograms forpCR

Van Stiphout R, Valntini V. et al. Radiother Oncol 2014

Van Stiphout R, Valentini V et al. Radiother Oncol 2011

Van Stiphout R, Valentini V et al. Radiother Oncol 2014

Nomogram

pCR (ypT0N0)

AUC = 0.70

highmediumlow

MRI ROIextraction FilterApplication

Dataanalysis(Moddicon):Modeldevelopment/validation

Radiomic inrectal cancer:

amodelforpCR prediction

Dinapoli N.et al.ESTRO35,Turin 2016

EXTERNAL

VALIDATION

Toremember

• Themeaning ofCR surg reduction• Howtoachieve CR RT;TIME• What toirradiate GTV-MR• Howtopredict nomograms

Maria Antonietta Gambacorta - Congressi · PDF fileAttualità nei trattamenti integrati...

Documents

Transcript of Maria Antonietta Gambacorta - Congressi · PDF fileAttualità nei trattamenti integrati...