Relapses after surgical treatment of ocular adnexal basal cell carcinomas: 5-year follow-up at the same university centre


Eckart Bertelmann
Department of Ophthalmology
Charité-University School of Medicine Berlin
Campus Virchow Klinikum
Augustenburger Platz 1
13353 Berlin
Tel: + 49 030 45055 4202
Fax: + 49 030 450 554900


Purpose:  To define the efficacy of surgical resection of ocular adnexal basal cell carcinoma and to analyse possible risk factors, we determined relapse rates on a yearly basis postintervention in a cohort of patients that were treated in our hospital under comparable conditions.

Methods:  A total of 366 such cases that were admitted for treatment between 2002 and 2006 were followed prospectively once yearly and tumour size and localization, histological type, functional and cosmetical outcome and relapses were recorded. In cases when the primary resection margins were not free, further resections were performed until the resection margin was histologically assessed tumour free on the basis of paraffin-fixed tissue sections. Lid reconstruction was carried out as direct closure, Tenzel rotation flap, tarsomarginal transplants or tarsoconjunctival flaps (Hughes), inversed rotation flaps, tarso marginal grafts, Cutler Beard flaps, free skin grafts, rhomboid, glabella or advancement flaps. Relapse-free survival 3, 4 and 5 years postoperatively was estimated by the Kaplan–Meyer method.

Results:  Three relapses after 3 years and two further relapses after 5 years were recorded. An additional relapse was documented 6 years postoperatively. Accordingly, the relapse-free survival was 0.99 (95% CI 0.98–1.00) after 3 years, 0.99 (95% CI 0.97–1.00) after 4 years and 0.97 (95% 0.95–0.98) after 5 years. The relapses occurred in previously relapsed tumours and in tumours of morphea type. Primary nodular basal cell carcinomas did not relapse in this study.

Conclusions:  Based on these studies and in comparison with published relapse rates after cryotherapy, surgical resection appears to be superior to the latter procedure.


Basal cell carcinoma is one of the most prevalent skin tumours in people with light complexion (Roewert-Huber et al. 2007), and its occurrence is still rising. In the United States and Australia, basal cell carcinoma is the most common malignancy, and in parts of Australia, the incidence is higher than 1% (Leibovitch et al. 2005). Reported incidences from European countries ranged from 0.5 to 3 per 100 000 inhabitants in Finland (Paavilainen et al. 2005) to 70 per 100 000 inhabitants in France (Allali et al. 2005). Well-known risk factors are rare genetic diseases like Golin-Goltz syndrome and most commonly exposure to UV-light (Brash 1997). Whereas squamous-cell carcinoma appears to be strongly related to cumulative sun exposure, the relationship between exposure to ultraviolet radiation and the risk of basal-cell carcinoma is more complex. The risk of this disease is significantly increased by recreational exposure to the sun during childhood and adolescence (Rubin et al. 2005). Moreover, exposition to higher UV doses seems to shift the ratio between basal cell and squamous cell carcinomas towards the more malignant squamous cell carcinomas (Ramos et al. 2004).

Treatment of ocular adnexal basal cell carcinomas includes various histologically controlled surgical techniques, moreover, radiotherapy, cryotherapy and curettage and electrodesiccation (Rodriguez-Vigil et al. 2007). In the last decade, new treatment options like photodynamic therapy (Wang et al. 1999), topical imiquimod (Leppälä et al. 2007) and interferon α2b (Fenton et al. 2002) have diversified the therapeutic arsenal. Thus, the evaluation of relapse rates and long-term prognosis of available methods is important for assessment of the most suitable treatment.

Commonly, surgery with micrographically controlled resection margins has been correlated with low recurrence rates (Morris et al. 2009). However, published relapse rates vary widely between different studies reaching from below 1% to over 4% after several years. (Lindgren et al. 2000; Malhotra et al. 2004; Hamada et al. 2005; Khandwala et al. 2005; Morris et al. 2009). Results of larger studies are shown in Table 1. Overall, the highest published relapse rates after surgical resection are comparable to those reported after cryotherapy, which has the advantage of lesser efforts of hospitalization. However, cancer recurrence 5 years after modern cryotherapy ranged from 3.5% up to 7.6% (Anders et al. 1995; Tuppurainen 1995; Buschmann 2002).

Table 1.   Clinical studies on recurrence rates of ocular adnexal basal cell carcinoma.
StudyDesignPatients includedLength of follow-up
Recurrence rate
Lindgren (2000)Prospective64495%
Hamada (2005)Retrospective223604.35%
Khandwala (2005)Retrospective93602.2% overall, 14.3% of recurrent tumours
Malhotra (2004)Prospective multi -centre819 (347 completed follow-up)602.0% overall 7.8% of recurrent tumours
Morris (2009)
Prospective278 (173 completed follow-up)600.58%

In the present study, we report on tumour recurrences in a cohort of 366 patients with surgically treated ocular adnexal cell carcinoma that were included in a prospective study from 1 January 2002 to 31 December 2006. In addition, we present our conclusions regarding possible risk factors that are involved.

Patients and Methods

In this prospective study, we included patients that were consecutively treated in our clinic for ocular adnexal basal cell carcinomas by the same surgeon from 2002 to 2006 and gave informed consent to be included in the study. Ocular adnexal basal cell carcinoma was defined as basal cell carcinoma in the lower eyelid, upper eyelid, medial and lateral canthus. The study period for inclusion of patients ran from 1 January 2002 to 31 December 2006. The study protocol was completely in accordance to the Declaration of Helsinki. Ten patients declined to enter the study because they feel not to be able to fulfil the postoperative control visits, so the enrolment ratio was 97%. All patients with basal cell carcinoma of the defined region (upper eyelid, lower eyelid, medial and lateral canthus) with primary or recurrent tumour were included as far as they agreed to enter the study. Patients with ocular adnexal malignancies of another entity were excluded. Moreover, patients treated permanently in a nursing home prior to first presentation in our institution were excluded because we estimated the chances low that they would be able to complete the follow-ups.

There were 366 patients (199 women, 167 men) of a mean age of 69 years (SD 8 years) who could be included in the study. Postoperative control visits were scheduled after 2 and 4 weeks, then 3 and 6 months thereafter. Subsequently, from 1 year after the resection on follow-up, inspections occurred on a yearly basic. Tumour size, localization, histological subtype, functional and cosmetical outcome and relapses were documented (Table 2).

Table 2.   Distribution of tumour localization, histological subtype and relation of primary and relapsed tumour in a cohort of 366 patients with ocular adnexal basal cell carcinoma.
LocalizationHistological subtypeNumber of tumour manifestations
Upper eyelid40Nodular150Primary301
Lower eyelid210Ulcerative120Relapsed65
Medial canthus90Morphea like96  
Lateral canthus26    

Nodular basal cell carcinomas were resected in a safety distance of 3 mm, previously relapsed basal cell carcinomas and morphea-like basal cell carcinomas were removed in a safety distance of at least 5 mm. Tumour samples were sent to histological examination after marking the resection margins. The histological assessment was performed on paraffin-fixed tissue sections. If one or more not-tumour-free resection margins were documented by the pathologist (61 patients, 16.7%), further resections were taken from the involved margin until the resection was tumour cell free. At the time point of wound closure/lid reconstruction in all patients, all resection margins were histologically assessed tumour free.

Lid reconstruction was carried out as direct closure, Tenzel rotation flap, tarsomarginal transplants or tarsoconjunctival flaps (Hughes) when the lower eyelid was involved. The upper eyelid was reconstructed by inversed rotation flaps, tarsomarginal grafts or Cutler Beard flaps. Free skin grafts, rhomboid flaps or glabella flaps were used for the medial canthus and free skin grafts or advancement flaps for the lateral canthus.

Cosmetic outcome was assessed as excellent, good, adequate or poor. The cosmetic result was judged as excellent if no visible scar could be seen in the patient’s face and the lid function was completely normal. The cosmesis was assessed as good if the lid position and function were completely normal but slight signs of the previous operation were still visible (patient in Fig. 4). Cosmesis was assessed as adequate if the lid position and function were normal but an operation scar was visible in the patients’ face or slight side differences of the lid position to the contralateral eye were visible (patient in Fig. 2). The cosmetic result of the operation was judged as poor when an operative revision was necessary (e.g. ectropium correction).

Figure 4.

 (A) Seventy-two year old woman at first presentation with morphea-like basal cell carcinoma of the right lower eye lid. Resection in 5 mm safety distance and reconstruction by a lateral rotation flap (Tenzel). (B) Relapse tumour 24 months after first resection. (C) Lower eye lid reconstruction by Hughes tarsoconjunctival flap and free skin graft.

Figure 2.

 (A) Seventy-nine year old woman at first presentation in our institution, relapsed basal cell carcinoma in the medial canthus. (B) Defect after resection in 7 mm safety distance. (C) Eighteen months later relapse at the lower margin of preceded localization 1st reconstruction with glabella flap. (D) Defect after resection of the 2nd relapse. (E) Four weeks after reconstruction by retroauricular skin graft. (F) One year after second reconstruction, well-healed skin graft, no further relapse.


Losses in follow-up are shown in Fig. 1 (below the x-axis). Overall, three relapses were documented after a postoperative follow-up of 36 months (18, 24 and 36 months). After 5 years, the number of relapses raised up to five patients. In the group of patients followed longer than 5 years, one further relapse occurred 6 years after the first operation. The Kaplan–Meyer estimate of the cumulative probability of remaining free of relapse is shown in Fig. 1. Accordingly, the relapse-free survival is 0.99 (95% CI 0.98–1.00) after 3 years of follow-up, 0.99 (95% CI 0.97–1.00) and 0.97 (95% CI 0.95–0.98) after 5 years of follow-up. Two of the patients with relapse after 3 years had a previous relapse before first presenting in our department (Fig. 2). Two of the relapses were located in the medial canthus region (Fig. 2). One relapse occurred after primary resection of a basal cell carcinoma of morphea-like growth type (Fig. 3). One of the two further relapses after 5 years, and the additional relapse after 6 years were involving the orbit (Fig. 4), one of these two patients, an 85-year-old patient with advanced Parkinson disease presented with a relapsed basal cell carcinoma when primarily treated in our clinic. The tumour consisted of a morphea-like basal cell carcinoma of the medial canthus involving the lacrimal fossa and the lacrimal sac. After resection with a safety margin of 5 mm and additional resection, the resection margins were tumour free and the defect was reconstructed with a glabella flap. When she presented with the relapsed tumour at the 4-year follow-up, the tumour mass had grown to the size of half of the globe. A biopsy confirmed the diagnosis of relapsed basal cell carcinoma. Because of her ill mental and general health, her family declined further treatment of the orbital tumour. The second patient with orbital basal cell carcinoma relapse presented initially with a morphea-like primary tumour involving the complete lower lid margin (Fig. 3). The tumour was resected with safety margin of 5 mm, and the lower eye lid was reconstructed with three free tarsomarginal grafts. At the examination 5 years later, the patient presented with a tumour mass in the lateral orbit with a diameter of 15 mm. The relapsed tumour was resected by an anterior orbitotomy. Then, the lateral canthus was reconstructed by a Tenzel rotation flap for the lower eye lid. In a subsequent operation of the upper eye lid, a Cutler–Beard flap, modified because of the preceding reconstruction of the lower lid, was then performed. In spite of good anatomical lid reconstruction with attaching lid margins and complete lid closure, the patient developed chronic surface alteration because of a severe dry eye symptom. The 6th patient with relapsed basal cell carcinoma had a morphea-like primary tumour of the mid-lower. After resection with 5 mm safety distance, the lid was reconstructed by a Tenzel rotation flap. The defect after resection of the relapsed tumour at the same position was reconstructed by a (Hughes) tarsoconjunctival flap (Fig. 4).

Figure 1.

 Kaplan–Meyer estimate of the cumulative probability of remaining free of relapse at different time points of follow-up in a cohort of initially 366 patients after surgical removal with histologically controlled margins of ocular adnexal basal cell carcinoma. Relapse-free survival is 0.99 (95% CI 0.98–1.00) after 3 years of follow-up (three relapses) and 0.97 (95% CI 0.95–0.98) after a follow-up of 5 years (two further relapses).

Figure 3.

 (A) Fifty-four year old man at first presentation with basal cell carcinoma of complete lower eyelid. (B) Resection in safety distance of 5 mm, lid reconstruction by three tarsomarginal grafts. (C) Orbital relapse in lateral canthus 6 years after first operation. (D) Lower lid reconstruction by rotation (Tenzel) flap, upper lid reconstruction by a modified Cutler Beard flap in a second procedure.

Over all, in three of our patients with relapsed basal cell carcinoma, a primary relapse had been treated at the first presentation in our clinic, four patients had a tumour of morphea-like growth type, and three of the relapsed tumours had been located in the medial canthus.

The cosmetic result was judged at the control 1 year postoperatively as excellent in 56%, good in 32%, adequate in 10% and poor in 2% of the patients.


In this study, we demonstrate 3–4 and 5-year results after surgical resection of ocular adnexal basal cell carcinoma. Comparison of 3- and 5-year relapse rates shows that additional relapses have to be expected even 3 years after primary resection. The overall percentage of relapses in our study is in the range of comparable studies.

The study confirms that surgical resection with safety distance and histologically controlled resection margins cannot guarantee a relapse-free follow-up after several years (Goldberg 1997; Lawrence 1999; Khandwala et al. 2005). The lowest recurrence rate after 5 years has been recently reported by Morris (Morris et al. 2009). The histological assessment of tumour-free resection margins in this study was guaranteed applying the slow Mohs technique.

Risk factors in our patients seem to be morphea-like subtype, preceding relapse and localization in the medial canthus. Primary nodular basal cell carcinomas did not relapse in our patient series. In a retrospective study in 2001, histological subtype and localization (medial or lateral canthus) were not confirmed as risk factors for tumour relapse (Zimmermann & Klauss 2001).

Theoretical reasons for incompletely controlled quantitative tumour removal include the following: insufficient safety distance, discontinuous tumour growth, incomplete histologically controlled resection margins because of discontinuously removed resection margins during surgery, incompletely prepared histological sections not representing the complete three-dimensional resection margin, and ignored tumour cells in the histological sections. Thus, the reasons may be related to surgery, histological assessment and tumour growth type itself.

In tumours of the morphea-like growth type, discontinuous tumour growth has been documented, and these tumours represented four of six relapsed tumours in our patient series. Moreover, discontinuous tumour growth can also occur in cases of previous tumour relapses (Lawrence 1999).

On the other hand, surgical resection of tumours growing in the medial canthus can be difficult. In tumours involving the lacrimal sac, the lacrimal system is included in the resection area. When the tumour reaches the periost, parts of it can be included in the resected tissue However, in the situation of deep tumour infiltration, the maintaining of a safety distance can be difficult, when the distance to the bone surface (medial orbital rim, lacrimal fossa) is less than 5 mm and bone removal is not included in the resection.

According to our views of the clinical consequences of our results, the safety distance in basal cell carcinoma of morphea-like growth type should be even greater than the clinically recommended 5–7 mm. Moreover, long-term (over 3 years) postoperative controls are essential for recognition and adequate treatment of possible relapses.