Recurrence patterns identify aggressive form of human papillomavirus‐dependent vulvar cancer

Vulvar cancer is rare and, as a result, is understudied. Treatment is predominantly surgery, irrespective of the type of vulvar cancer, and is associated with physical, emotional and sexual complications. A cluster of human papillomavirus (HPV)‐dependent vulvar cancer patients was identified in Arnhem Land Northern Territory (NT), Australia, in which young Indigenous women were diagnosed at 70 times the national incidence rate.


INTRODUCTION
In 2009, a cluster of vulvar cancer, an otherwise rare disease, was identified in Arnhem Land, a remote and sparsely populated region of the Northern Territory (NT), Australia. 1 Indigenous women aged <50 years had an age-adjusted incidence rate of vulvar cancer of 31.1 per 100 000 (95% CI 13.1-49.1), or 70 times the national Australian rate of 0.4 per 100 000 (95% CI 0.4-0.5) in this age group. Investigation of this cluster found all tested cases to be positive for human papillomavirus (HPV), but that the increased incidence could not be explained by increased infection with high-risk HPV types or infection with a particularly virulent variant of HPV16. 2,3 This suggests the involvement of another cofactor, such as an environmental exposure or genetic predisposition. While a preliminary genetic investigation found no effect of genome-wide homozygosity or any individual region of homozygosity on vulvar squamous cell carcinoma (VSCC) and vulvar intraepithelial neoplasia (VIN) in the East Arnhem cluster, the possibility of a genetic risk factor has yet to be eliminated. 4 Vulvar cancer has received sparse attention in the research literature, in part because it is rare (3-5% of all gynaecological cancers). 5,6 VSCC comprises the majority (>90%) of vulvar cancers. 7 There are two distinct aetiological pathways for VSCC, resulting in separate forms of the disease. [7][8][9] One is associated with infection with HPV, usually affects younger women, arises from vulvar high-grade squamous intraepithelial lesions (HSIL) (previously called usual type VIN (uVIN)) 10 and results in basaloid or warty carcinoma. The other variant is HPV-independent, usually affects postmenopausal women, is associated with differentiated VIN (dVIN) arising in an area affected by vulvar dermatoses such as lichen sclerosus, and leads to keratinising VSCC. Evidence to date indicates that dVIN progresses to invasive VSCC more frequently and faster than HSIL, and is more likely to recur after treatment. 9,11 Vulvar neoplasms are predominantly treated with surgery, irrespective of whether the disease is HPV-dependent or independent.
Treatment for VSCC and VIN (both HSIL and dVIN) has evolved only marginally and continues to be associated with substantial physical morbidity and psychosexual dysfunction. 12,13 A trend toward less radical surgery and an increase in the use of sentinel lymph node biopsies has gone some way toward reducing morbidity for patients with localised disease without reducing survival. 12,14 Options for treating advanced or recurrent VSCC remain limited, contributing to higher mortality rates in these patients. 15,16 Mortality rates have remained steady, or increased in some regions, and women suffer from associated morbidities that reduce quality of life. [17][18][19] Greater understanding of the development of VSCC and its precursor lesions, VIN, is needed to improve preventive or therapeutic strategies.
Recurrence of VSCC occurs in 12-37% of patients, with most recurrences (40-80%) occurring within two years of treatment. 20 Rates of recurrence have remained steady over the past 30 years, and the causes of local recurrences are poorly understood, contributing to a lack of preventive and therapeutic options. 21 Risk factors previously identified as associated with recurrence include smoking, larger lesion size, inadequate surgical excision margins, stromal invasion, and treatment with laser ablation. 20,22,23 However, more recent studies indicate that these risk factors, and especially positive margins, are less influential in determining recurrence than the presence of tumour-adjacent epithelium already molecularly altered by inflammation (eg chronic lichen sclerosus) or infection (ie high-risk HPV), 21 and the small numbers involved in most studies mean that current evidence of risk factors for recurrence remains equivocal. 24 Gynaecologists treating vulvar cancer in the NT over the past 20 years have noted that women from East Arnhem are more likely to suffer recurrences than other women with vulvar cancer.
This study therefore assesses whether women from the Arnhem Land cluster differ from women with VSCC and VIN resident elsewhere in the NT in recurrence after treatment, disease progression and mortality.

Data sources
Cases and recurrences were ascertained from the Colposcopy Database maintained by the Gynaecology Outreach Service (GOS) and the Royal Darwin Hospital, which comprises records of all colposcopies performed by the GOS and public gynaecology services since 1996. This was supplemented by the NT Cancer Registry and information from client medical records in the NT public hospitals' clinical information system, allowing the dataset to be extended back to January 1993.

Definitions
Women were included if they were NT residents diagnosed with VIN or VSCC between 1 January 1993 and 30 June 2015; women were excluded if they were diagnosed with VIN or VSCC but had no record of treatment, recurrence, or any further information.
The data sources used do not distinguish between HSIL and dVIN; consequently, VIN is used to refer to all high-grade vulvar intraepithelial neoplasias.
The index diagnosis for cohort inclusion was histologically con-

Statistical methods
Data for the cohort are summarised as: for categorical variables, frequency and percentage; for continuous variables, mean and standard deviation or median and interquartile range. Differences between groups for categorical variables were analysed using χ 2 or Fisher's exact test, where applicable. Time to recurrence is summarised for Indigenous and non-Indigenous women using cumulative incidence plots; these are preferred over Kaplan-Meier when there is the presence of a non-negligible competing risk, in this case mortality. 25 Fine and Gray competing risk regression models were therefore employed to account for this competing risk in estimating the sub-distribution hazard of recurrence for Indigenous versus non-Indigenous women adjusted for age at di-  (Table 1). Recurrence was more likely in younger women, those initially diagnosed with VIN, and women from East Arnhem. The rate of recurrence in women from Central Australia was comparatively low.
The cumulative incidence of recurrence ( Fig. 1)  Recurrence is more frequent in Indigenous women, compared to non-Indigenous women (Fig. 2). At five years after diagnosis,   Checks. While a tendency to present with more advanced disease may contribute to the higher fatality rate observed in affected Arnhem Land women in the years immediately following treatment, it is notable that those women diagnosed initially with VIN were more likely to experience recurrence after treatment than women diagnosed with cancer.
The higher mortality rate observed in Indigenous women compared to non-Indigenous women at five years disappears by 10 years. This is explained by the fact that this is all-cause mortality (data were not available for cause-specific mortality) and non-Indigenous women are on average almost 10 years older than Indigenous women at age of diagnosis. It is therefore likely that difference in mortality between the two groups is eliminated by 10 years because all-cause mortality increases in non-Indigenous women due to older age. While it would be optimal to report these separately, the numbers  Our findings provide support for: increased resources for HPV vaccination in Arnhem Land, to ensure maximal three-dose cover- that are both more targeted and associated with less morbidity and mortality than current treatments. 37