Oncological outcome and long-term complications in robot-assisted radical surgery for early stage cervical cancer: an observational cohort study

Authors

  • JP Hoogendam,

    Corresponding author
    1. Department of Gynaecological Oncology, Woman and Baby Division, University Medical Centre Utrecht, Utrecht, the Netherlands
    • Correspondence: Dr JP Hoogendam, Department of Gynaecological Oncology, Woman and Baby Division, University Medical Centre Utrecht, Utrecht, F05.126, PO Box 3508, 3508 GA, the Netherlands. Email j.hoogendam@umcutrecht.nl

    Search for more papers by this author
  • RHM Verheijen,

    1. Department of Gynaecological Oncology, Woman and Baby Division, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • I Wegner,

    1. Department of Gynaecological Oncology, Woman and Baby Division, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • RP Zweemer

    1. Department of Gynaecological Oncology, Woman and Baby Division, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author

Abstract

Objective

To report the oncological outcome and long-term complications of radical surgery by robot-assisted laparoscopy in early stage cervical cancer.

Design

Observational cohort study.

Setting

Tertiary referral centre.

Population

About 100 cervical cancer patients treated consecutively with robot-assisted radical surgery between 2008 and 2013.

Methods

Two gynaecological oncologists specialised in minimally invasive surgery performed all surgeries on a three/four-armed robotic system. Procedures consisted of pelvic lymph node dissection combined with a radical hysterectomy, radical vaginal trachelectomy or parametrectomy.

Main outcome measures

Recurrence, survival and long-term complication rates.

Results

104 robot-assisted laparoscopies were performed in 100 patients (stage IA1–IIB), with a median follow-up of 29.5 months (range 2.5–67.1 months). Thirteen cases were diagnosed with a loco-regional (8%), distant (4%) or combined (1%) recurrence at a median of 14.4 months (range 2.9–34.8 months). All mortality (7%) was cervical cancer-related and due to recurrent disease. Four recurrences receive palliative care and two are in complete remission. The overall 5-year progression-free and disease-specific survival rates are 81.4 and 88.7%, respectively. Frequent complications were lymphoedema (26%), lower urinary tract symptoms (19%), urinary tract infection (17%) and sexual disorders (9%). Five patients had a vaginal cuff dehiscence. No complication-related mortality occurred.

Conclusion

The recurrence, survival and long-term complication rates of robot-assisted radical surgery for early stage cervical cancer in this cohort are reassuring concerning its continued clinical use.

Introduction

In early stage cervical cancer, pelvic lymph node dissection (PLND) combined with a radical hysterectomy is regarded as the primary treatment of choice. In select cases, a radical trachelectomy can successfully preserve fertility.[1] A sentinel node (SN) procedure can be included and is aimed at ascertaining lymph nodal (LN) status at the beginning of surgery.[2, 3]

Disease recurrence rates of 9–18% after an open PLND and radical hysterectomy have been reported for stage IB1-IIA cervical cancer.[4-6] In addition, the long-term complications include lymphedema (4–41%), sexual complaints (19–35%), lower urinary tract symptoms (LUTS; 13–42%) and urinary tract infections (2–49%).[4, 7-13]

In recent years, robot-assisted laparoscopy has increasingly been adopted and facilitates a minimally invasive approach in cervical cancer patients.[14] Robot-assisted surgery allows for better cosmetic results, reduces postoperative pain and intraoperative blood loss, shortens recovery and the time of hospital admissions.[15-17] However, this is counterbalanced by substantial setup costs, increased overall duration of surgical procedures, bulky, expensive equipment with absent tactile feedback and learning curve issues.[18-20]

While both the positive and negative immediate effects have been abundantly discussed, only very limited data are available on the long-term oncological outcomes and complications in robotically treated patients.[21] Such information is urgently needed about this relatively new and rapidly expanding surgical modality.

This study reports the oncological outcome and complications of currently the largest single institutional cohort of consecutive cases treated by robot-assisted laparoscopy for early stage cervical cancer.

Methods

Research population

We performed an observational cohort study on the first 100 cervical cancer patients treated consecutively between February 2008 and July 2013 at our tertiary referral centre. Subjects were included when primary cervical cancer was histopathologically proven and at least one session of robot-assisted laparoscopic surgery was performed. Cases were retrieved from the departmental registry for surgery in gynaecological oncology. All procedures were part of the standard clinical care, for which routine informed consent was obtained.

Surgery

As part of standard care, high-risk stage IA1-2 and IB1/IIA patients underwent a robot-assisted laparoscopic SN procedure, PLND and radical surgery. Treatment consisted of radical hysterectomy, or radical vaginal trachelectomy when fertility preservation was desired and the maximum tumour diameter was ≤2 cm. In patients with an anticipated risk of >10% for LN metastasis, only a robot-assisted laparoscopic SN procedure and PLND was performed. Radical surgery followed in a second session after histopathological confirmation of negative LN status. In some cases, robotic PLND and parametrectomy expanded an earlier simple hysterectomy with an unsuspected finding of carcinoma.

Within a trial conducted by the European Organization for Research and Treatment of Cancer (EORTC 55994), patients with stage IB2, IIA or IIB cervical cancer received neo-adjuvant chemotherapy. After confirmation of an adequate response, a robot-assisted laparoscopic PLND and radical hysterectomy were performed.[22]

Two gynaecological oncologists (R.V., R.Z.) specialised in minimally invasive surgery performed all robot-assisted gynaecological laparoscopies. Their robotic experience before the first case included in this study, was three (R.V.) and two (R.Z.) years.

All surgeries were performed under general anaesthesia with a three- or four-armed robotic surgical system (da Vinci®; Intuitive Surgical, Sunnyvale, CA, USA). Routine antibiotic prophylaxis consisted of 2000 mg cefazolin and 500 mg metronidazole. In dorsolithotomy, an urinary catheter was inserted. After skin disinfection and draping, a pneumoperitoneum of 24 mmHg was created using a supra-umbilically inserted Verres needle. Following insufflation, a zero degree robotic laparoscope was introduced via an 11-mm trocar. Two robotic trocars (8 mm) were positioned 8 cm bilaterally of the umbilicus and a fourth was placed cranial to the right iliac crest. A conventional trocar (12 mm) was positioned lateral to Palmer's point.[23] The robot was docked and surgical instruments were introduced in a 28° Trendelenburg position and at a reduced pneumoperitoneal pressure of 12–14 mmHg.

Routinely, a SN procedure precedes the PLND at our institution. At 17 hours preoperatively, 220–290 MBq technetium-99m nanocolloid is intrastromally injected into four peritumour quadrants with subsequent SN mapping by either lymphoscintigraphy or SPECT-CT. Directly prior to peritoneal incision, a four-quadrant injection through a pre-fixed system with 0.5–1.0 ml patent-blue dye (Bleu Patenté, Guerbet, Roissy, France) was given to further enable both visual and gamma probe-assisted SN identification. Resected SNs were examined intraoperatively by frozen section and postoperatively by multiple sectioning and immunohistological staining. In patients with tumour-positive SNs, the radical hysterectomy was substituted by chemoradiation. Details on this methodology have been published recently.[24]

The PLND involved resection of all lymphoid tissue located in the obturator fossa and surrounding the internal, external and common iliac vessels. Typically, a Querleu–Morrow type C1 radical hysterectomy was performed with a level two lymphadenectomy.[25] Uterine manipulation during surgery was performed with a 35/45 mm McCartney tube (LiNA Medical, Glostrup, Denmark).

Postoperative care was performed on a standard care ward by staff specialised in oncological care. The urinary catheter was removed upon adequate mobilisation and followed by two ultrasonographic measurements of the post-void residual urine volume. Postoperative thrombophylaxis consisted of daily subcutaneous injections with 5700 IU nadroparin.

Data collection

All individual information on follow-up was retrospectively retrieved from the institutional medical records. Follow-up at our centre entails ambulant visits to a gynaecological oncologist at intermissions of 3 months (1st year), 4 months (2nd year) and 6 months during the last 3 years. At each visit, an interval history and full gynaecological examination are completed. Additional diagnostic tests or imaging was performed only when clinically indicated. Disease recurrence was defined as the presence of cancer tissue originating from the primary tumour, following a follow-up period in which the disease status was considered ‘no evidence of disease’.

A complication was defined as any undesired and (potentially) harmful event to the patient's wellbeing which might have been related to the robot-assisted laparoscopic procedure. The adjective ‘long-term’ refers to those complications detected or persisting during the oncological follow-up programme. Complications were scored based on the Common Terminology Criteria for Adverse Events (CTCAE) guideline version 4.03.[26] Outcomes of complications were scored at the last available follow-up date and were categorised as complete, partial, no recovery or death.

Statistical analysis

Statistical calculations were performed with the Statistical Package for the Social Sciences version 20.0.0 (SPSS; International Business Machines, Armonk, NY USA). The analysis of oncological outcomes was performed case-wise and the complications on an operation-wise basis. Cases wherein robot-assisted surgery was discontinued were included in the analysis (i.e. intention-to-treat analysis). Depending on the data type and its distribution, appropriate summary statistics and (non-)parametric tests were used. Statistical significance was set at P < 0.05, producing confidence intervals (CI) at the 95% level.

For survival data, a Kaplan–Meier analysis was performed with log-rank testing for survival differences. Survival is described as the time interval between the end of primary treatment and the moment of disease recurrence, detected clinically, by imaging or by histopathological biopsy (progression-free survival), or until death from cervical cancer (disease-specific survival) or death from any cause (overall survival).

Results

Population

A total of 104 robot-assisted laparoscopies were performed in 100 patients (Table 1): by year, 10 (2008), 13 (2009), 24 (2010), 21 (2011), 19 (2012) and 13 (2013). Seven cases had received chemotherapy as part of the on-going EORTC 55994 trial prior to surgery.[22] All robotic surgeries were completed except for one case (1%) in which a robotic PLND was finished by conventional laparoscopy due to technical difficulties. A combined SN procedure, PLND and radical hysterectomy was the most frequent (= 58) surgery with a median procedure length (i.e. skin incision to closure) of 319 minutes (range 175–472 minutes). For this procedure, the median intraoperative blood loss was 185 ml (range 50–750 ml) and no blood transfusions were required. Discharge followed after a median of four postoperative days (range 2–8 days). The histopathological characteristics are outlined in Table 2.

Table 1. The baseline characteristics for the study population (= 100). Percentages are omitted, as they would equal n-values
Median age (range)41 (23–81) years
Median BMI (range)23.2 (18.3–35.1) kg/m2
 No. of patients
  1. ASA, American Society of Anaesthesiologists; BMI, body mass index; FIGO, International Federation for Gynecology and Obstetrics; LLETZ, large loop excision of the transformation zone; PLND, pelvic lymph node dissection; RH, radical hysterectomy; RVT, radical vaginal trachelectomy; SN, sentinel node.

  2. a

    Treated surgically following sufficient response to neo-adjuvant chemotherapy in a trial setting (EORTC 55994).[22]

  3. b

    Includes four patients who underwent PLND (±SN) and RH in two separate robotic sessions.

Smoking 27
Diabetes mellitus 3
Hypertension 11
Prior abdominal surgery
Open surgery15
Laparoscopic surgery10
Route unspecified7
Prior cervical procedure
LLETZ41
Conisation16
Parity
030
1–245
≥319
Missing6
ASA classification
171
227
Missing2
FIGO stage
IA1-27
IB183
IB2a4
IIA4
IIBa2
Robotic surgery
SN + PLND + RH58
SN + PLND + RVT7
SN + PLND10
PLND + RH7
PLND + Parametrectomy8
Otherb10
Table 2. Histopathological results of the study population (= 100)
Median tumour diameter (range)18 (3–43) mm
Median no. LN resected (range)a24 (10–56)
  n %
  1. LN, lymph nodes; PLND, pelvic lymph node dissection.

  2. a

    Applicable in cases undergoing PLND (= 98).

  3. b

    Applicable in cases undergoing resection of the parametria (= 88).

Tumour histology
Squamous cell carcinoma6868
Adenocarcinoma2727
Other55
Tumour differentiation grade
I1717
II5656
III2323
Undefined44
Lymph-vascular space invasion 4646
Parametrial invasion b 78
LN status a
Macrometastases88
Micrometastases22
Isolated tumour cells66

There were no cases with tumour-positive resection margins. Twenty-three cases received chemoradiation or adjuvant radiotherapy because of resection margins <5 mm (= 6), parametrial invasion (= 4), LN micro- or macrometastases (= 10), or a combination of these factors (= 3). Of the 10 patients with LN metastases, seven were found to be tumour-positive at frozen section and three at final histology.

Oncological outcomes

The mortality corrected median follow-up duration was 29.5 months (range 2.5–67.1 months). Thirteen cases (13%) had a recurrence, of which eight were loco-regional, four distant and one presented with a combined recurrence (Table S1). The median follow-up time until recurrence detection was 14.4 months (range 2.9–34.8 months). One subject presented with an umbilical trocar site metastasis. Ten recurrences were symptomatic, of which nine presented outside of the follow-up program. Three asymptomatic recurrences were discovered during a regular follow-up visit at 7, 9 and 31 months post-surgery. Two patients (2%) developed a second primary gynaecological malignancy (micro-invasive vaginal carcinoma and ovarian carcinoma stage IIIC).

All mortality (= 7) was cervical cancer related. All seven cases died of recurrent disease and another four are currently receiving palliative care. Two patients with a recurrence, isolated at the vaginal cuff, entered complete remission following (chemo) radiation. The overall (= 100) 5-year progression free and disease specific survival were 81.4% and 88.7% (Figure 1). For stage IB1 (= 83) these rates were 80.5% and 87.9%, respectively.

Figure 1.

(A) Kaplan–Meier curve illustrating the progression-free survival of the research population (= 100). (B) Corresponding curve for the disease-specific survival. The overall survival curve is omitted as all mortality was cervical cancer-related.

The 5-year progression-free and disease-specific survival for cases with a tumour-negative LN status was 87.4 and 95.0%, respectively. In the subgroup with tumour-positive LN, these rates were significantly (both < 0.001) lower, at 27.4 and 35.0%, respectively (Figure 2).

Figure 2.

(A) Kaplan–Meier curve illustrating the progression-free survival for patients with a tumour-negative (blue) or tumour-positive (green) lymph nodal status. (B) The corresponding disease-specific survival.

Complications

Intraoperative complications occurred in 12 cases (12%), of which bladder (= 4) and ureter (= 3) injuries were the most prevalent. One subject had severe subcutaneous emphysema and required prolonged ventilation. Two complications were anaesthesia-related (bronchospasm, supraventricular tachycardia) and another two patients had a systemic allergic reaction to patent blue dye. All made a complete recovery and no conversions to a laparotomy were performed.

In the PLND population (= 98), 25 patients (26%) reported lymphedema for any length of time or severity during follow-up. Twenty received non-invasive therapy and five were monitored expectantly, resulting in complete (= 8, 32%), partial (= 3, 12%) or no recovery (= 14, 56%). The presence of lymphedema was not significantly influenced by adjuvant treatment (= 0.556) or by the number of resected LN (= 0.150). Nine patients were diagnosed with a lymphocele, of which four were clinically relevant due to infection (all made a complete recovery) (Table S2).

Of the 89 cases with (para)uterine surgery, 24 patients (27%) required urinary self-catheterisation following clinical discharge. Persisting LUTS (>6 weeks postoperatively) were present in 17 (19%), of which eight required intermittent self-catheterisation for any length of time, resulting in complete (= 7, 41%), partial (= 3, 18%) or no recovery at the time of last follow-up (= 7, 41%) of LUTS. The relative risk (RR) of urinary catheterisation at discharge of developing LUTS was elevated, at RR 5.0 (CI 2.1–11.9). Seventeen patients in the (para)uterine surgery subgroup (19%) had ≥1 urinary tract infection (UTI). The RRs of catheterisation at discharge and LUTS of developing a UTI were RR 6.5 (CI 2.6–16.4) and RR 3.8 (CI 1.7–8.3), respectively.

In the (para)uterine surgery subgroup, eight patients (9%) reported dyspareunia, a lubrication disorder and/or reduced libido. Five patients (6%) had a vaginal cuff dehiscence, of which four were superficial (mucosal tearing only) and did not require surgery. Of all surgically treated patients, 11 patients (11%) experienced iatrogenic menopause, seven of whom opted for hormone replacement therapy. Four cases (4%) experienced hypoaesthesia of skin areas innervated by the obturator or genitofemoral nerve.

Six cases required surgery because of a complication: haemorrhaging (= 2, 2%), ureterovaginal fistula (= 2, 2%), abscess (= 1, 1%) or vaginal dehiscence (= 1, 1%). This overlaps with the 17 patients who were readmitted for a median of 5 days (range 1–9 days). The diagnoses for readmission were an infected lymphocele (= 4, 4%), complicated UTI (= 3, 3%), infected haematoma (= 3, 3%), abscess (= 2, 2%), fistula (= 2, 2%) or other (= 3, 3%). No complication led to mortality (Table S2).

Discussion

Main findings

We present the oncological outcomes and complication rates of the largest single institution, robotically treated cervical cancer cohort published to date. Our results suggest that overall recurrence and survival rates are comparable to those achieved with non-robotic treatment in the Netherlands (5-year survival stage IB1 = 88%, IIA = 76%),[27] especially when taking into account that we included patients with initial stage IB2, IIA2 and IIB disease after neo-adjuvant chemotherapy. Furthermore, the results show that a robot-assisted laparoscopic SN procedure is feasible and that the LN status it ascertains, remains the principal prognostic factor in this population.

In our cohort, LN-negative patients (= 90, 90%) had an excellent prognosis of a 95.0% 5-year survival, which is similar to larger studies of open surgery.[28] We encountered only 10 patients with positive LNs (= 10, 10%), which had an exceptionally poor prognosis. Five patients died from recurrent disease. In four of these, all treated with chemoradiation after PLND, the recurrence was distant (Table S1: cases 7, 8, 11 and 13). These deaths are therefore unlikely to be related to the surgical modality used. Previous studies focusing on recurrence after laparoscopic treatment for cervical cancer have found equal or improved recurrence rates compared with open surgery.[29] However, no firm conclusions can be drawn from our small numbers or from the relatively small earlier studies, and further research is needed to investigate any possible relationship between (robot-assisted) laparoscopic surgery and recurrence localisation, especially in LN-positive patients.

Strengths and limitations

Some limitations of this study merit further explanation. During the 5.5-year period studied, parallel cervical cancer surgeries via laparotomy or conventional laparoscopy were only incidentally performed at our institution. Hence, confounding by indication (i.e. selection bias) is not likely to have influenced our results. A second type of bias is the potential misclassification of the outcome status (i.e. information bias), which is an inherent danger of observational studies. This entails missing a long-term complication because it is not reported by the patient (e.g. sexual problems) or adequately registered by the physician. Here, our limited median follow-up needs to be taken into account.

This study does not intend – nor is it designed – to make any inference concerning the correlation between the surgeon's experience and the occurrence of complications or oncological outcomes. First, we consider it a study strength that both surgeons were already advanced in their learning curve prior to the first included subject.[15] Secondly, the true robotic proficiency of the surgeons is under-represented by our focus on cervical cancer. Throughout the inclusion period, both surgeons regularly performed robot-assisted laparoscopies for other indications. Thirdly, any comparison between observational studies on complication rates is hindered by non-uniform definitions of complications.

Interpretation

Currently, there is no published level A evidence comparing robotic versus open radical cervical cancer surgery on oncological outcomes for early stage cervical cancer. However, one randomised controlled trial (= 740) is currently underway and is primarily aimed at testing the equivalence to the 4.5-year disease-free survival rate. The secondary endpoints of the trial conducted by Obermair et al. include intra-and postoperative complications, quality of life, cost-effectiveness and the feasibility of the SN procedure.[30] The latter is especially interesting as both robot-assisted laparoscopy and the SN procedure have independently gained clinical acceptance in the past 5 years. When combined, a personalised minimally invasive surgical approach can be achieved.[24] Until these data are available (estimated: 2017), our study offers an initial insight into the long-term outcomes of radical robotic surgery performed in conjunction with an SN procedure for early stage cervical cancer.

Cantrell et al.[31] compared a robotic population (= 63) with an historical laparotomy cohort and demonstrated overall and progression-free survival of 94% for the robotic radical hysterectomy group (median follow-up 12 months). In the absence of any significant survival differences between both cohorts, they stated that patients may be preoperatively counselled to expect equivalent oncological outcomes.[31] In a smaller study (= 35), yet with a longer follow-up period (median 33 months), robotic radical cervical cancer surgery had a recurrence rate of 14.2%. While a formal survival analysis was omitted, at the time of publication one recurrence subject had died and the remaining four were still alive.[32] In contrast, various other small studies (= 13–30) have reported low recurrence rates of 0–4.3% in their radically treated cervical cancer cohorts (follow-up 12, 12, 25 and 31 months).[17, 21, 33, 34]

Clearly, published evidence on oncological safety is at a disadvantage due to small cohort size or limited follow-up time, which is the result of the relatively recent implementation of robotic surgery.

In our robotically treated cohort, long-term complications were frequently encountered and had substantial morbidity. This can be expected, as complications (e.g. lymphedema) depend predominantly on the procedure (e.g. PLND) and less on the technique (laparoscopy versus laparotomy). Consequently, the prevalence of such complications cannot be expected to diminish substantially, regardless of the surgical modality chosen.

We experienced a relatively high prevalence of intraoperative urinary tract injuries (7%). Lower urinary tract symptoms (19%) were also more frequent when compared with the 7–13% found in other robot-assisted radical hysterectomy studies specifying various bladder symptoms.[16, 19, 33] Besides our inclusion of a wider spectrum of (subjective) urinary complaints, inter-study differences in follow-up length and baseline risks may also have contributed to this discrepancy. The same holds true for lymphoedema in our study (26%), which was encountered more often than the 2–13% reported in robotic literature.[16, 20, 23, 33] This might be (partly) explained by our practice of including every patient who had (self-)reported oedema, regardless of its severity. The presence of very mild cases can be concluded from the fact that a significant number of affected patients did not require compressive stockings or physical therapy.

Vaginal cuff dehiscences are more frequently reported after laparoscopic hysterectomies and, according to a recent overview, especially after robot-assisted procedures (3.2%).[35] This percentage is comparable to our cohort, with five dehiscences, although four only involved superficial opening of the vaginal cuff.

Conclusion

The recurrence and survival rates of the presented cohort appear equivalent to the rates accepted for open surgery in early stage cervical cancer. However, observational results such as these offer only an initial insight; ultimately, randomised controlled trials are needed to definitively establish the oncological safety of robot-assisted laparoscopic surgery. Although previously demonstrated short-term benefits are reproduced, our results indicate that long-term complications remain substantial. Future studies focusing on surgical techniques disregarded in this study (e.g. nerve-sparing strategies) could uncover additional benefits of robotic surgery. In view of all these considerations, and in the absence of conclusive level A evidence, we support the continued use and research in robotic surgery for early stage cervical cancer.

Disclosure of interests

The authors declare they have no conflicts of interest related to the presented research.

Contribution to authorship

J.P.H., R.H.M.V. and R.P.Z. designed the study. R.H.M.V. and R.P.Z. performed all robotic surgeries. Data collection was carried out by J.P.H. and I.W. The statistical analysis was done by J.P.H. and the subsequent results interpreted by J.P.H., R.H.M.V. and R.P.Z. J.P.H. and R.P.Z. drafted the manuscript, which was critically revised and approved by all the authors.

Details of ethics approval

This study was based on an audit of the departmental complication and treatment outcome register, which is maintained as a part of standard clinical care and primarily aims to improve that care. No ethical review board approval is required for the use of such data.

Funding

None.

Acknowledgements

The authors are grateful to Ina Jürgenliemk-Schulz, MD, PhD, (radiation oncologist, UMC Utrecht) and Petronella O. Witteveen, MD, PhD (medical oncologist, UMC Utrecht) for their contributions to this study and manuscript.

Ancillary