Patients undergoing lymphadenectomy for stage III melanomas of known or unknown primary site do not...

Patients undergoing lymphadenectomy for Stage III melanomasof known or unknown primary site do not differ in outcome

Maria Celia Hughes1*, Annaliesa Wright2*, Andrew Barbour3, Janine Thomas3, B. Mark Smithers3,

Adele C. Green1,4 and Kiarash Khosrotehrani2

1 Queensland Institute of Medical Research, Cancer and Population Studies group, Brisbane, QLD, Australia2 The University of Queensland, UQ Centre for Clinical Research, Experimental Dermatology Group, Brisbane, QLD, Australia3 Queensland Melanoma Project, Discipline of Surgery, The University of Queensland, Princess Alexandra Hospital, Woolloongabba, QLD, Australia4 Department of Dermatological Sciences, School of Translational Medicine, University of Manchester, Manchester Academic Health Sciences Centre,

Manchester, United Kingdom

The outcome of patients with palpable melanoma metastases in lymph nodes in the presence (metastatic melanoma of known

primary site, MKP) or absence (metastatic melanoma of unknown primary site, MUP) of an identifiable primary tumour remains

controversial. Some of the previous studies contained large case series that included historical patients. We aimed to com-

pare outcomes of those with MUPs versus MKPs with palpable lymph node invasion, after staging with modern imaging tech-

nology. Aprospective study of patients from a single tertiary institution who were undergoing lymph node dissection for

palpable metastatic melanoma between 2000 and 2011 was conducted. All patients were ascertained by computerised tomog-

raphy scanning and most diagnosed after 2004 had positron emission tomography scanning also. Clinicopathological details

about the primary melanoma and lymph node dissections were gathered. Factors associated with recurrence and melanoma-

specific mortality in those with MKP and with MUP were assessed using univariate and multivariate analyses. Out of 485

patients studied, 82 had MUP and 403 had MKP. Patients were followed up for a median of 17.4 and 19.0 months, for MKP

and MUP, respectively. Five-year adjusted melanoma-specific survival was 58% for MUPs versus 49% for MKPs and was not

significantly different between the two groups (adjusted Cox proportional Hazard ratio 5 0.88 95% confidence interval [0.58,

1.33] p 5 0.54). Previously established prognostic factors such as number of positive nodes and extracapsular extension

were confirmed in both sets of patients. We conclude that among melanoma patients presenting with clinically detectable

nodes, when accurately staged, those without an identifiable primary lesion have similar outcomes to patients with MKP.

In a material proportion of cases, melanoma metastases arediagnosed without any identifiable primary tumour. Melanomawith unknown primary lesion (MUP) is defined as melanomaconfirmed histologically and following the exclusion of muco-sal melanoma.1 Patients presenting with MUP represent 2–3%of all melanoma patients.1–4 Population-based studies in theUnited States and The Netherlands have reported that 2.2 and2.6% of all melanomas were of unknown primary.3,5

The outcome of patients with MUP has been the subjectof debate. Despite their new AJCC classification,6 MUPs withlymph node involvement represent a heterogeneous popula-tion in terms of outcome. When compared to metastatic mel-anoma of known primary site (MKP), an improved overalland disease-free survival has been reported.2,7,8 However,these findings have been contradicted by others.3,9,10 Animportant consideration is the role of historical data in com-paring MUPs versus MKPs. Complete workup and accuratestaging of melanoma patients with lymph node involvementis essential to determine further management.2,7,10,11, Overthe last three decades, the accuracy of staging has changedlargely with the introduction of computerised tomography(CT) and positron emission tomography (PET) scanners.12

These techniques have become part of the routine only in thepast decade, potentially changing the staging of patients con-sidered to have palpable lymph node disease only, by moreaccurately confirming or refuting distant disease.13 This moreaccurate staging obviously affects reported recurrence andsurvival potentially affecting comparisons between MUPs andMKPs in large series using historical data.

In our study, we aimed to compare the outcome of a con-temporary cohort of patients (diagnosed 2000–2011) withMUP versus MKP who presented with palpable lymph nodes

Key words: melanoma, metastasis, unknown primary, prognosis,

survival

Additional Supporting Information may be found in the online

version of this article.

*M.H. and A.W. contributed equally to this work

Grant sponsor: NHMRC Program; Grant number: 552429; Grant

sponsor: NHMRC career development fellowship; Grant number:

1023371

DOI: 10.1002/ijc.28318

History: Received 23 Dec 2012; Accepted 7 May 2013; Online 10

Jun 2013

Correspondence to: Kiarash Khosrotehrani, The University of

Queensland Centre for Clinical Research, Royal Brisbane Hospital

Building 71/918, Herston, QLD 4029, Australia, Tel: 161733466077,

Fax:161733465598, E-mail: [email protected]

ShortReport

Int. J. Cancer: 133, 3000–3007 (2013) VC 2013 UICC

International Journal of Cancer

IJC

and who were confirmed to be Stages IIIB and IIIC usingmodern staging techniques.

Subjects and MethodsPatients referred to the Princess Alexandra Hospital Mela-

noma Clinic for regional lymphadenectomy (LND) for meta-static melanoma between January 2000 and June 2011 wereconsidered. Demographic data, operative details and survivaloutcomes were documented prospectively from 2004 and allcases of nodal surgery were reviewed retrospectively to theyear 2000 and maintained on a database as approved by theMetro South Health Service District Human Research EthicsCommittee (HREC/2004/QPAH/018). For our study, patientspresenting between January 2000 and June 2011 were deemedeligible if they (i) had palpable regional lymph-node metasta-sis of melanoma; (ii) had complete cervical, axillary or ingui-nal LND; (iii) were aged >18 years; (iv) were staged usingCT scan of the head, chest, abdomen and pelvis. From 2004,the use of FDG-PET scan increased and has become moreroutine in the majority of patients. When considered rele-vant, directed scans such as bone scans and MRI of the brainhave also been used. (v) Patients did not show any sign ofvisceral, skin or distant lymphatic metastases.

Patients who had a positive sentinel lymph node or whohad radiotherapy prior to LND were not included. Adjuvantsystemic therapy after LND including interferon and radio-therapy was permitted.

Histological characteristics of primary (where relevant)and nodal metastatic melanoma were collected prospectivelyuntil August 31, 2011. For primary melanomas, the site(head and neck, arm, leg and trunk), thickness of tumour (inmm), histological classification, the presence of ulceration,satellites or regression were extracted. Patients were consid-ered as having a MUP if a primary melanoma could not beclinically identified on the skin or the mucosa. For LNDs, thedissection site (neck, axilla, iliac, ilio-inguinal and inguinal),total nodes removed, number of involved nodes (number ofpositive nodes), size of largest node (in cm) and the presenceof micro or gross extra-capsular extension were recorded.Post-surgery complications such as infection and seromarequiring active intervention, by aspiration or formal drain-age, were also noted.

Patients were followed prospectively by clinical examina-tion every 3 months for the first 2 years, every 6 months foryears 3 and 4, and then annually up to 10 years. Investigation

and imaging was performed when considered clinically indi-cated. Recurrence was defined by the identification of local,regional or disseminated disease and the date was recorded.Date and cause of death were collected and confirmed fromthe records of the Queensland Cancer Registry and the Regis-ter of Births, Deaths and Marriages, Queensland, for allevents occurring until December 31, 2011.

Our study was approved by the Human Research EthicsCommittees of the University of Queensland, the PrincessAlexandra Hospital and the Queensland Institute of MedicalResearch.

Outcomes

Comparing patients with MUP and MKP, the two out-comes assessed were (i) risk of disease recurrence and (ii)risk of death due to melanoma. Time to recurrence was cal-culated from date of dissection to date of first recurrence,whereas time to death was calculated from date of dissectionto date of death. Patients who did not have a recurrence bythe time of their last follow-up date were censored at thattime for analysis of risk of recurrence. For analysis of risk ofdeath, patients were censored if no death was recorded byDecember 31, 2011. Patients who died prior to December 31,2011 due to causes other than melanoma were censored atdate of death.

Statistical analysis

We used univariate and multivariate Cox Proportional Haz-ard regression models to estimate Hazard ratios (HRs) com-paring risk of recurrence and melanoma death in patientswith MUPs to patients with MKPs. For multivariate analyses,possible confounding factors were selected among the perso-nal characteristics of patients, histological characteristics oftheir metastatic melanoma and post-surgery complications intwo stages. In the first stage, we assessed factors associatedwith recurrence and melanoma death using univariate CoxProportional Hazard regression models and all variables stat-istically significant at the 5% level were then included in apreliminary multivariate model that always included MUP/MKP as a variable. In the second stage, we fitted into thepreliminary multivariate model the variables that were origi-nally not significant in the univariate models to identify fac-tors that were detectable only after adjusting for major riskfactors. Factors that remained significant or changed risk esti-mates by 10% or more were retained in the final models.

What’s new?

Whether the outcome of patients presenting with palpable melanoma metastases in lymph nodes is better in the absence of

an identifiable primary tumour remains controversial. Most studies were based on historical cases with potential staging bias.

Here the authors conducted a large cohort study among stage III melanomas with palpable disease of known or unknown

primary tumour after proper staging using modern imaging technology. While some predictors of recurrence or death

previously described in stage III melanoma patients were confirmed, the lack of identifiable primary melanoma had no

significant impact on patient outcomes in terms of recurrence and melanoma-specific death.

ShortReport

Hughes et al. 3001


Finally, Kaplan–Meier survival curves showing the numbersof patients at risk at various follow-up points and exact timesof death and recurrence were established. We then used esti-mates from the final adjusted models and applied the cor-rected group prognosis method to produce adjustedmelanoma-specific and disease-free survival curves comparingMUPs to MKPs.14,15 In the corrected group prognosismethod, the coefficients from the final survival model usingthe entire study sample are used to calculate individual sur-vival curves for all combinations of variables in the model.The average survival is then calculated as a weighted averageof the individual curves, with weights proportional to thenumber of individuals at each level of the covariates. Todetermine if risk of recurrence and melanoma death differedby the number of positive nodes, analyses were repeated forthose with 1, 2–3 and 4 or more invaded nodes, separately.

Differences in age and sex of patients and histologicalcharacteristics of primary and metastatic melanoma betweenMKPs and MUPs were assessed using chi-square tests forcategorical or nominal variables, t-test for age and non-parametric methods (Wilcoxon–Mann–Whitney) for all otherquantitative variables. All analyses were performed using Sta-tistical Analysis Software (SAS) version 9.2 (SAS Institute,Cary, NC) with added relevant macro.16

ResultsFrom January 2000 to June 2011, 573 patients underwentLND. From this group, 485 were eligible for our study asthey presented with a palpable lymph node. Other cases wereexcluded as they had only microscopic lymph node involve-ment or sometimes Stage IV disease. Among eligible patients,403 (83%) had a MKP and 82 (17%) had MUP. Patientsaffected with MKP or MUP did not differ in terms of age,sex or dissection site (Table 1). Similarly, the number of posi-tive nodes was not different between MKP and MUP. How-ever, patients with MUPs were more likely to have largernodes without any differences in extra-capsular extension orpost-surgical complications (Table 1). Overall, MUPs weremore likely to be staged IIIB versus IIIC when compared toMKPs at study entry.

MKPs and MUPs were followed for a median of 17.4(interquartile range [IQR] 5 31.8, range: 0.2–135.6) and 19.0months (IQR 5 34.9, range: 0.6–134.0) respectively (non-sig-nificant). Disease recurrence was identified in 212 cases over-all (43.7%), 181 among MKPs and 31 among MUPs. Localrecurrence occurred in 23 (10.8%) with no difference betweenMUP and MKP patients (10.5 vs. 12.9%, p 5 0.69). Mediantime to recurrence for 181 MKPs was 7.4 months (IQR 5

8.3, range: 0.4–103.1 months) and for 31 MUPs, 7.7 months(IQR 5 10.8, range: 1.4–31.1 months). Two hundred andnine patients (43.1%) died during follow-up with melanomalisted as a cause of death in 199 patients (172 MKPs and 27MUPs). Median time to death for MKPs was 13.9 months(IQR 5 18.5, range: 0.4–90.5 months) as compared to 16.5(IQR 5 14.4) months for MUPs (range, 3.9–41.8).

MUPs as compared to MKPs had 17 and 25% lower riskof recurrence and melanoma-specific death (non-significant)(Table 2 and Supporting Information Fig. 1). Increasingnumber of positive nodes, the presence of micro- or grossextra-capsular extension, Stage IIIC staging and the presenceof a post-surgery seroma requiring intervention were all posi-tively associated with recurrence and were included in thepreliminary model for multivariate analysis. Having a groindissection (inguinal or ilio-inguinal dissection) as well asincreasing number of harvested nodes became significantlyassociated with recurrence only after adjustment for theabove major risk factors. As the number of nodes that can beharvested is dependent on the site of dissection, this was notincluded in the final model. After full adjustment, the differ-ence in risk of disease recurrence between MUPs and MKPsremained not significant (adjusted HR: 0.81; 95% confidenceinterval [CI]: [0.55, 1.19], p 5 0.28). Five-year adjustedrecurrence-free survival was 50% for MUPs versus 44% forMKPs (Fig. 1a). Adjusted analysis restricted to groups ofMUPs and MKPs with only 1, 2–3 or >4 nodes producedsimilar results (Supporting Information Table 1).

Regarding melanoma-specific mortality, age at LND of751 years, increasing number of positive nodes, the presenceof micro- or gross extra-capsular extension, Stage IIIC diseaseand post-surgery seroma were positively associated with mel-anoma death. No other risk factors were identified in the sec-ond stage of analysis. After adjustment for the above riskfactors, MUPs and MKPs were not different for the risk ofdeath due to melanoma (adjusted HR: 0.88 95% CI [0.58,1.33], p 5 0.54 and Fig. 1b). Five-year adjusted melanoma-specific survival was 58% for MUPs versus 49% for MKPs aswas overall survival: 58% for MUPs and 49% for MKPs basedon all available patient data. This difference was not signifi-cant (adjusted HR 5 0.86 [0.58, 1.28], p 5 0.47). Adjustedanalysis restricted to groups of MUPs and MKPs with only 1,2–3 or >4 nodes produced similar results (Supporting Infor-mation Table 2).

DiscussionPatients presenting with node-positive disease without thepresence of a known cutaneous or mucosal primary mela-noma represent a heterogeneous population and often raisedifficult management questions. The outcome in terms ofrecurrence or survival in this group of patients has been con-troversial. We here compared the outcomes of MUP com-pared to MKP while appropriately taking into account otherprognostic factors, in a recently ascertained cohort of patientsafter LND for melanoma. Our cohort study did confirm pre-dictors of recurrence or death previously described in StageIII melanoma patients. However, we found that the lack ofidentifiable primary melanoma had no significant impact onpatient outcomes in terms of recurrence and melanoma-specific death in univariate or multivariate analyses.

Several studies have compared MUPs to MKPs amongStage III patients with metastatic melanoma. Lee et al.7 found

ShortReport

3002 Survival in melanomas of unknown primary


Table 1. Demographic, clinical and histological characteristics of melanoma patients with a known (MKP) or unknown (MUP) primary site

Characteristic

MKP (n 5 403) MUP (n 5 82)

n1 % n1 % p-Value

Sex

Male 256 63.5 57 69.5

Female 147 36.5 25 30.5 0.30

Age at LND (years)

<35 42 10.4 9 11.0

35 to <45 50 12.4 12 14.6

45 to <55 88 21.8 13 15.9

55 to <65 67 16.6 20 24.4

65 to <75 76 18.9 17 20.7

751 80 19.9 11 13.4 0.37

Age at surgery (years) (mean 6 sd) 57.5 6 16.6 56.8 6 16.9 0.74

Site of primary melanoma

Head and neck 73 18.1 – –

Arm 45 11.2 – –

Leg 133 33.0 – –

Trunk 152 37.7 – – NA

Breslow thickness

T1: 0.2–1.0 mm 89 22.7 – –

T2: >1.0–2.0 148 37.8 – –

T3: >2.0–4.0 92 23.5 – –

T4: >4.0 63 16.1 – – NA

Histology

SSM 254 63.0 – –

NM 100 24.8 – –

LMM 7 1.7 – –

Other 23 5.7 – –

Missing 19 4.7 – – NA

Ulceration

No 279 70.3 – –

Yes 118 29.7 – – NA

Satellites

No 387 98.0 – –

Yes 8 2.0 – – NA

Regression

No 355 90.6 – –

Yes 37 9.4 – – NA

Dissection site

Neck 98 24.3 27 32.9

Axilla 156 38.7 26 31.7

Iliac 4 1.0 3 3.7

Ilio-inguinal 28 7.0 6 7.3

Inguinal 117 29.0 20 24.4 0.15

ShortReport

Hughes et al. 3003


a significant survival benefit in MUP patients, with 5-yearoverall survival rates of 55% for MUP versus 44% for MKP.This represented an HR of overall survival at 1.5. Subgroupanalysis revealed that the survival difference was significantonly for those with more than one invaded lymph node. Sim-ilarly, Cormier et al.2 reported 5-year overall survival rates of55% for MUP versus 42% for MKP. Prens et al.8 demon-strated 5-year overall survival rates of 40 and 27% for MUPand MKP, respectively. In our study, MUPs and MKPs didnot differ in terms of melanoma outcome. We also

performed subgroup analyses and found that in the MUPgroup the presence of more than two positive lymph nodeshad the same hazard for death and recurrence when com-pared to MKPs. Although we found a trend towards bettersurvival of MUP patients, this might only reflect the fact thata larger number of them were staged as IIIB as compared toMKPs on inclusion, potentially giving this group a survivaladvantage. Our study was sufficiently powered to detect a dif-ference as great as that described previously.8 We cannot for-mally exclude that in a much larger hypothetical cohort the

Table 1. Demographic, clinical and histological characteristics of melanoma patients with a known (MKP) or unknown (MUP) primary site(Continued)

Characteristic

MKP (n 5 403) MUP (n 5 82)

n1 % n1 % p-Value

Number of nodes removed

1–11 105 26.1 15 18.3

12–18 100 24.8 24 29.3

19–27 101 25.1 21 25.6

281 97 24.1 22 26.8 0.49

Number of nodes removed(median, Q1, Q3)

18 (11, 27) 19 (12, 29) 0.62

AJCC staging

N1: 1 201 49.9 48 58.5

N2: 2–3 125 31.0 19 23.2

N3: 41 77 19.1 15 18.3 0.30

Number of positive nodes(median, Q1, Q3)

2 (1,3) 1 (1, 2) 0.23

Size of largest node

0.2–2 cm 102 25.4 12 14.8

>2.0–3.0 128 31.8 23 28.4

>3.0–4.0 93 23.1 16 19.8

>4.0 79 19.7 30 37.0 0.005

Size of largest node (cm)(median, Q1,Q3)

2.9 (2.0, 4.0) 3.5 (2.3, 5.0) 0.007

Extra capsular extension

Nil 216 53.6 43 52.4

Micro 168 41.7 34 41.5

Gross 19 4.7 5 6.1 0.87

Tumour stage

IIIB 199 49.4 55 67.1

IIIC 204 50.6 27 32.9 0.004

Infection

No 344 85.4 74 90.2

Yes 59 14.6 8 9.8 0.24

Seroma

No 324 80.4 67 81.7

Yes 79 19.6 15 18.3 0.78

1For some characteristics, summed total may be less than number of patients due to missing values.Q1, Q3: value of first and third quartiles.

ShortReport



Table 2. Personal and histological characteristics associated with recurrence and melanoma death: univariable hazard ratios (95% CI)

Characteristic

Recurrence Melanoma death

HR 95% CI p-ValueOverallp-value HR 95% CI p-Value

Overallp-value

MUP

No 1 – 1 –

Yes 0.83 0.56, 1.21 0.32 0.32 0.75 0.50, 1.13 0.17 0.17

Sex

Male 1 – 1 –

Female 0.98 0.74. 1.30 0.90 0.91 0.93 0.69, 1.25 0.62 0.62

Age at LND (years)

<35 1 – 1 –

35 to <45 1.60 0.91, 2.81 0.11 1.32 0.74, 2.33 0.35

45 to <55 1.17 0.69, 2.00 0.56 0.89 0.52, 1.54 0.68

55 to <65 0.91 0.51, 1.62 0.75 0.92 0.52, 1.63 0.78

65 to <75 1.17 0.68, 2.02 0.57 0.94 0.54, 1.64 0.83

751 1.40 0.81, 2.42 0.22 0.25 1.76 1.05, 2.96 0.03 0.01

Site of primary melanoma

Head and neck 1 – 1 –

Arm 1.07 0.61, 1.89 0.81 1.20 0.68, 2.12 0.54

Leg 1.34 0.87, 2.07 0.18 1.13 0.72, 1.78 0.59

Trunk 1.06 0.69, 1.63 0.80 0.44 1.21 0.78, 1.86 0.39 0.86

Breslow thickness

T1: 0.2 to 1.0 mm 1 – 1 –

T2: >1.0–2.0 1.00 0.67, 1.49 0.99 1.13 0.75, 1.72 0.56

T3: >2.0–4.0 1.49 0.97, 2.27 0.07 1.55 1.00, 2.40 0.05

T4: >4.0 1.01 0.62, 1.65 0.97 0.14 1.02 0.60, 1.73 0.95 0.17

Histology

SSM 1 – 1 –

NM 1.17 0.84, 1.62 0.36 1.23 0.88, 1.72 0.23

LMM 0.22 0.03, 1.54 0.13 1.23 0.45, 3.53 0.68

Other 1.03 0.56, 1.92 0.92 1.10 0.57, 2.10 0.78

Missing 0.28 0.09, 0.87 0.03 0.08 0.79 0.37, 1.70 0.55 0.70

Ulceration

No 1 – 1 –

Yes 1.01 0.73, 1.39 0.96 0.96 1.19 0.87, 1.64 0.28 0.28

Satellites

No 1 – 1 –

Yes 1.33 0.50, 3.59 0.57 0.57 2.19 0.90, 5.34 0.08 0.08

Regression

No 1 – 1 –

Yes 0.87 0.52, 1.45 0.58 0.58 1.31 0.82, 2.09 0.26 0.26

Dissection site

Neck 1 – 1 –

Axilla 1.05 0.74, 1.49 0.80 1.11 0.77, 1.58 0.58

Iliac 1.86 0.74, 4.66 0.19 1.14 0.36, 3.65 0.83

ShortReport

Hughes et al. 3005


difference in survival between MUP and MKP such as weobserved might have been statistically significant. We note,however, that the difference we observed in a recently ascer-tained and carefully characterised cohort of Stage IIIB andIIIC melanoma is much smaller than that reported in theprevious studies, making MUP less important clinically thanpreviously supposed.

Our Melanoma Unit is a major referral centre for the stateof Queensland, making a selection bias improbable. Giventhe similar management strategies in all studies assessing theoutcome of patients with MUP versus MKP, the explanationfor the different conclusions may relate to the method ofstaging at inclusion of patients. Previous studies included his-torical cases from the 1980s to recent times.2,7,8 We have

focused on a more recent and narrow window of prospectiverecruitment: data collection started from the year 2000 whenall patients were staged with a CT scan and many with FDG-PET scanning as opposed to chest X-rays and abdominalultrasound.13 Our cohort is therefore more likely to be devoidof patients with Stage IV disease who may have been inap-propriately considered in historical series. This might also bereflected in the slightly better overall 5-year survival that wereport compared to other studies.

In accordance with our results, similar-to-slightly betterrates of survival have been reported by Rutkowski et al.9

when comparing outcomes in MUP versus MKP. A recentstudy by de Waal et al.3 found no significant difference insurvival rates when comparing MKP with MUP patients with

Table 2. Personal and histological characteristics associated with recurrence and melanoma death: univariable hazard ratios (95% CI)(Continued)

Characteristic

Recurrence Melanoma death

HR 95% CI p-ValueOverallp-value HR 95% CI p-Value

Overallp-value

Ilio-inguinal 1.47 0.80, 2.71 0.21 0.93 0.46, 1.90 0.85

Inguinal 1.35 0.94, 1.93 0.11 0.26 1.20 0.82, 1.74 0.34 0.89

Number of nodes removed

1–11 1 – 1 –

12–18 0.74 0.51, 1.07 0.11 1.05 0.72, 1.55 0.80

19–27 0.75 0.51, 1.08 0.12 0.92 0.62, 1.36 0.66

281 0.67 0.46, 0.98 0.04 0.16 0.90 0.60, 1.34 0.60 0.84

AJCC staging

N1: 1 positive node 1 – 1 –

N2: 2–3 1.75 1.28, 2.41 0.0005 1.83 1.31, 2.54 0.0003

N3: 41 3.26 2.33, 4.57 <0.0001 <0.0001 3.42 2.42, 4.83 <0.0001 <0.0001

Size of largest node

0.2–2 cm 1 – 1 –

>2.0–3.0 1.08 0.75, 1.56 0.68 1.59 1.06, 2.39 0.02

>3.0–4.0 1.34 0.91, 1.97 0.14 1.66 1.08, 2.55 0.02

>4.0 1.05 0.69, 1.58 0.82 0.47 1.60 1.03, 2.51 0.04 0.08

Extra capsular extension

Nil 1 – 1 –

Micro 1.89 1.43, 2.51 <0.0001 2.17 1.61, 2.91 <0.0001

Gross 2.40 1.36, 4.23 0.002 <0.0001 3.73 2.23, 6.24 <0.0001 <0.0001

Stage

IIIB 1 – 1 –

IIIC 1.70 1.29, 2.23 0.0001 0.0001 2.00 1.50, 2.66 <0.0001 <0.0001

Infection

No 1 – 1 –

Yes 1.04 0.71, 1.52 0.84 0.84 1.10 0.75, 1.62 0.64 0.64

Seroma

No 1 – 1 –

Yes 1.76 1.30, 2.39 0.0003 0.0003 1.82 1.32, 2.50 0.0002 0.0002

ShortReport



macroscopic node involvement. It is worth noting our studyalso recruited patients between 2003 and 2009 and was morelikely to offer a more precise staging. This heterogeneity infindings clearly suggests that once a lymph node has beeninvaded by melanoma, the presence or absence of a knownprimary lesion or the characteristics of the primary in generalhave minor or no influence on the course of the disease, asfound in our study.

ConclusionsIn conclusion, after accurately staging patients with palpablemetastatic melanoma to regional lymph node, we show that

the difference in survival between MUPs and MKPs wassmaller than previously reported and did not reach signifi-cance. Although more powered studies may, in the future,report this small difference as valid, we propose that thisparameter may not be clinically relevant for the managementof patients with Stage III melanoma.

AcknowledgementsThe authors thank Vic Siskind for helpful discussion. M.Hughes was supported by NHMRC Program Grant 552429.K Khosrotehrani was supported by NHMRC career develop-ment fellowship 1023371.

References

1. Pfeil AF, Leiter U, Buettner PG, et al. Melanomaof unknown primary is correctly classified by theAJCC melanoma classification from 2009. Mela-noma Res 2011;21:228–34.

2. Cormier JN, Xing Y, Feng L, et al. Metastatic mel-anoma to lymph nodes in patients with unknownprimary sites. Cancer 2006;106:2012–20.

3. de Waal AC, Aben KK, van Rossum MM, et al.Melanoma of unknown primary origin: apopulation-based study in the Netherlands. Eur JCancer 2013;49:676–83.

4. Katz KA, Jonasch E, Hodi FS, et al. Melanoma ofunknown primary: experience at MassachusettsGeneral Hospital and Dana-Farber Cancer Insti-tute. Melanoma Res 2005;15:77–82.

5. Chang AE, Karnell LH, Menck HR. The NationalCancer Data Base report on cutaneous and non-cutaneous melanoma: a summary of 84,836 casesfrom the past decade. The American College ofSurgeons Commission on Cancer and the Ameri-can Cancer Society. Cancer 1998;83:1664–78.

6. Balch CM, Gershenwald JE, Soong SJ, et al. Finalversion of 2009 AJCC melanoma staging andclassification. J Clin Oncol 2009;27:6199–206.

7. Lee CC, Faries MB, Wanek LA, et al. Improvedsurvival after lymphadenectomy for nodal metas-tasis from an unknown primary melanoma. JClin Oncol 2008;26:535–41.

8. Prens SP, van dPAP, van AAC, et al. Outcomeafter therapeutic lymph node dissection inpatients with unknown primary melanoma site.Ann Surg Oncol 2011;18:3586–92.

9. Rutkowski P, Nowecki ZI, Dziewirski W, et al. Mela-noma without a detectable primary site with metasta-ses to lymph nodes. Dermatol Surg 2010;36:868–76.

10. Schlagenhauff B, Stroebel W, Ellwanger U, et al.Metastatic melanoma of unknown primary originshows prognostic similarities to regional meta-static melanoma: recommendations for initialstaging examinations. Cancer 1997;80:60–5.

11. Anbari KK, Schuchter LM, Bucky LP, et al. Mela-noma of unknown primary site: presentation,

treatment, and prognosis—a single institutionstudy. University of Pennsylvania PigmentedLesion Study Group. Cancer 1997;79:1816–21.

12. Bastiaannet E, Uyl-de Groot CA, Brouwers AH,et al. Cost-effectiveness of adding FDG-PET orCT to the diagnostic work-up of patients withstage III melanoma. Ann Surg 2012;255:771–6.

13. Bastiaannet E, Oyen WJ, Meijer S, et al. Impactof [18F]fluorodeoxyglucose positron emissiontomography on surgical management of mela-noma patients. Br J Surg 2006;93:243–9.

14. Ghali WA, Quan H, Brant R, et al., InvestigatorsA. Comparison of 2 methods for calculatingadjusted survival curves from proportional haz-ards models. J Am Med Assoc 2001;286:1494–7.

15. .Nieto FJ, Coresh J. Adjusting survival curves forconfounders: a review and a new method. Am JEpidemiol 1996;143:1059–68.

16. Liu L, Forman S, Barton B. Fitting cox modelusing PROC PHREG and beyond in SAS. SASGlobal Forum 2009. 2009;Paper 236-2009.

Figure 1. Recurrence-free and melanoma-specific survival of MUP versus MKP patients with palpable lymph node metastasis. (a)

Recurrence-free and (b) Melanoma-specific adjusted survival curves estimated using the corrected group prognosis curves for MUPs versus

MKPs.

ShortReport

Hughes et al. 3007


Patients undergoing lymphadenectomy for stage III melanomas of known or unknown primary site do not...

Documents

Transcript of Patients undergoing lymphadenectomy for stage III melanomas of known or unknown primary site do not...