The role of adjuvant radiotherapy in the local management of desmoplastic melanoma




In the current study, the authors sought to evaluate outcomes, specifically with respect to adjuvant radiotherapy (RT), for patients with desmoplastic melanoma.


The records of 130 consecutive patients who presented between 1985 and 2009 with nonmetastatic desmoplastic melanoma and were treated curatively with either surgery alone (59 patients; 45%) or surgery and postoperative RT (71 patients; 55%) were retrospectively reviewed. Ages ranged from 21 years to 97 years (median age, 66 years). The location of the primary tumor was in the head and neck region in 62% of patients. Only 5 patients (4%) had lymph node involvement at the time of presentation.


The median follow-up was 6.6 years (range, 11 months-24 years). Overall survival rates at 5 years and 10 years were 69% and 53%, respectively. Disease-specific survival rates were 84% and 80%, respectively, at 5 years and 10 years. The actuarial rate of local recurrence was 17% at 5 years and beyond. Of the patients who underwent surgery without receiving postoperative RT, 14 (24%) experienced local recurrence. Of the 71 patients treated with surgery and postoperative RT, 5 (7%) experienced local recurrence. In a Cox multivariate regression model, improved local control was significantly associated with the receipt of postoperative RT (P = .009).


Surgery followed by postoperative RT appears to provide superior local control compared with surgery alone for patients with desmoplastic melanoma. Cancer 2014;120:1361–1368. © 2013 American Cancer Society.


Desmoplastic melanoma is relatively rare variant of malignant melanoma that exhibits distinct clinical behavior from nondesmoplastic melanoma. The overall prognosis for patients with desmoplastic melanoma is believed to be favorable compared with patients with conventional melanoma subtypes in that distant metastases and regional lymph node metastases are relatively infrequent compared with nondesmoplastic melanoma.[1-3] However, some investigators have suggested that local recurrence (LR) rates for desmoplastic melanoma are higher than for conventional melanoma[4-6] and correspondingly emphasized the importance of aggressive local therapy in determining the overall disease-specific outcomes for patients with this malignancy.[7, 8] The role of radiotherapy (RT) as an adjuvant to wide local excision remains uncertain. Although it has been suggested that wide local excision is adequate to eradicate localized desmoplastic melanoma,[9] many reports have suggested that adjuvant RT improves local control for this disease.[3, 4, 10, 11]

Given the relative rarity of desmoplastic melanoma, all of the published literature for this disease to the best of our knowledge consists of retrospective reviews of clinical outcomes and potential prognostic factors. The data are conflicting with regard to whether neurotropism, a feature that is common although not universal in patients with desmoplastic melanoma, portends an increased risk of local failure. Similarly, a head and neck location of the primary tumor has been shown to be a potential poor prognostic factor, which is of particular relevance because these tumors commonly occur in a head and neck location. Finally, of particular recent interest is the prognostic extent of desmoplasia in the tumor (ie, whether the tumor is pure desmoplastic melanoma or whether the extent of desmoplasia occurs as some component of an otherwise nondesmoplastic melanoma lesion [combined/mixed desmoplastic melanoma]).[12, 13] Investigators have reported on various implications of the extent of desmoplasia with regard to whether it affects the risks of LR or disease-specific death. We undertook a review of our clinical experience at the University of Texas MD Anderson Cancer Center (MDACC) for patients with localized desmoplastic melanoma who were treated definitively with either surgery or surgery followed by postoperative irradiation. The goal of the current study was to report on outcomes, specifically with respect to adjuvant RT, as well as prognostic factors for patients with this disease.


Between 1985 and 2009, 130 consecutive patients with nonmetastatic desmoplastic melanoma presented for definitive local management at MDACC. Review of the medical records and RT charts that constitute the data for the current study commenced after approval was obtained from our Institutional Review Board. Patients underwent a full history and physical examination, routine blood tests, and appropriate imaging before receiving treatment. A histologic diagnosis of desmoplastic melanoma was confirmed in each case through a review of the pathology slides by a pathologist at MDACC at the time of presentation.

Patient and Tumor Characteristics

Of the patients in this study, 85 (65%) were men and 45 (35%) were women ranging in age from 21 years to 97 years (median, 66 years). The anatomic distribution of the primary lesions were as follows: 29 (22%) on the scalp, 44 (34%) on the face, 7 (5%) on the neck, 19 (15%) on the trunk, 26 (20%) on the upper extremity, and 5 (4%) on the lower extremity. Thus, 62% of the patients presented with a head and neck presentation of desmoplastic melanoma.

Twenty of the 130 patients (15%) presented to MDACC with a recurrent presentation of their disease, with 13 patients having undergone 1 prior surgery, 6 patients having had 2 prior surgeries, and 1 patient having had 3 prior surgeries for their disease. None of the patients who developed disease recurrence had received prior RT. Of the 117 patients for whom information regarding Clark level was available, 44 patients (38%) had Clark level IV disease and 73 (62%) had Clark level V disease. American Joint Commission on Cancer[14] T classifications were as follows: 1 patient (1%) with T1 disease, 15 patients (13%) with T2 disease, 32 patients (27%) with T3 disease, and 69 patients (59%) with T4 disease. The median tumor thickness of the primary lesion was 5.1 mm (range, 1 mm-30 mm).

Patients were designated as having pure desmoplastic melanoma (100 patients; 77%) if their pathology report indicated that there were no other invasive melanoma histologic subtypes in the specimen. Patients whose pathology report indicated that their melanoma contained only a desmoplastic component or desmoplastic features in the presence of other histologic subtypes were coded as having combined/mixed desmoplastic melanoma (30 patients; 23%). For the current study, the extent of desmoplasia as a percentage of the specimen could not be retrospectively determined in all patients. Information regarding perineural invasion was available for 101 patients and of those, 66 (65%) exhibited perineural invasion. Of these, 5 patients presented with named nerve invasion. Of those 89 patients for whom the presence of ulceration was evaluated, it was found to be present in 9 patients (10%).


The decision to use adjuvant RT for patients with desmoplastic melanoma was at the discretion of the treating surgeon and radiation oncologist for each patient and practice patterns varied among practitioners over the years encompassed by this study. Management of the primary lesion was accomplished with surgery alone in 59 patients (45%) and surgery followed by postoperative RT in 71 patients (55%). Surgery was as follows: wide local excision in 122 patients (94%), excisional biopsy in 2 patients, total rhinectomy in 2 patients, lower lip excision and segmental mandibulectomy in 2 patients, amputation of a toe in 1 patient, and wide local excision plus orbital exenteration in 1 patient. After definitive surgery, 121 patients (93%) had negative resection margins and 9 patients (7%) had positive or uncertain resection margins. There was no significant correlation noted between the presence of positive or uncertain resection margins and the receipt of postoperative RT (P = .413). Specifically, 3 of the 9 patients with positive/uncertain resection margins received postoperative RT and 6 did not. Eighty-three patients (64%) underwent a plastic surgery closure procedure as part of their definitive surgical management.

Assessment of regional lymphatics was as follows. A total of 64 patients were found to have no clinical evidence of lymph node involvement, and the lymph node regions were subsequently observed; and 56 patients underwent sentinel lymph node biopsy, none of whom were found to have lymph node involvement. Two patients underwent excisional biopsy of a clinically suspicious lymph node mass, one of which was confirmed to be metastatic disease. Three patients underwent selective lymphadenectomy, one of who had confirmed lymph node disease at the time of presentation. Five patients underwent full dissection of the regional lymphatics, 3 of whom were confirmed to have lymph node involvement at the time of diagnosis. Only 5 patients (4%) were found to have lymph node involvement at the time of presentation, 4 with cervical lymph node involvement and 1 with involvement of the inguinal lymph node basin. Of those with lymph node involvement, 4 were among the patients with pure desmoplastic melanoma and 1 had mixed desmoplastic melanoma.

Of the 71 patients who received RT, 68 patients were treated with 30 Gray (Gy) in 5 fractions delivered twice weekly (on a Monday/Thursday or Tuesday/Friday schedule) over 2.5 weeks. Two patients received 36 Gy in twice-weekly fractions of 6 Gy per fraction. One patient was treated with 60 Gy in 30 fractions of 200 centigray per fraction. The median total RT dose was 30 Gy (range, 30 Gy-60 Gy) and the median fractional dose was 6 Gy per fraction (range, 2 Gy-6 Gy per fraction). Treatment techniques included appositional electron fields in 52 patients (73%), photon fields in 13 patients (18%), matched photon and electron fields in 5 patients (7%), and intensity-modulated RT in 1 patient (1%). A margin of 3 cm to 4 cm around the surgical site was attempted when possible. Occasionally, wider fields were used to treat regional nerves. Of the 71 patients who received RT to the primary tumor site, 18 (25%) also received RT to the draining lymph node region.

Follow-Up and Statistical Analysis

Follow-up was calculated from the date of surgical resection. The median follow-up of those patients who were still alive at the time of last follow-up was 6.6 years (range, 11 months-24 years).

The Kaplan-Meier method[15] was used to calculate the actuarial curves for overall survival (OS), disease-specific survival (DSS), local control, distant metastatic recurrence, disease-free survival (DFS), and complication rates. The log-rank statistic was used to test for the significance of differences between the curves. Multivariate regression analysis was performed using the Cox proportional hazards model.[15] Differences between percentages for categorical variables were analyzed using the chi-square statistic or the Fisher exact test as appropriate. Surgical complications were defined as complications from surgery that could arise within 3 months of the date of surgery (regardless of whether the patient received RT) or that were explicitly deemed a surgical complication by a surgeon. Surgical and RT-related complications were scored as mild (self-limiting and requiring no treatment), moderate (requiring conservative medical management), and severe (requiring surgical intervention or hospitalization.



At time of last follow-up, 53 patients had died (41%) with a median survival of 11.8 years. A total of 23 patients (18%) died of their desmoplastic melanoma. The actuarial OS rates at 5 years and 10 years were 69% and 53%, respectively. The DSS rates were 84% and 80%, respectively, at 5 years and 10 years and the DFS rates were 72% and 70%, respectively, at 5 years and 10 years. The last recurrence occurred at 8.7 years from the date of surgical resection. Figure 1 shows the curves for OS, DSS, and DFS. Table 1 shows the univariate DSS. Lymph node involvement at the time of diagnosis and positive/uncertain resection margins were found to be adversely prognostic for DSS (Table 1). On the multivariate analysis, lymph node involvement continued to predict a poorer DSS (P < .0001) as did positive/uncertain resection margins (P = .03), even when adjusting for receipt of postoperative RT.

Table 1. Univariate Analysis for DSS
CharacteristicNo. (% of Total)10-Year DSS RateP
  1. Abbreviations: DSS, disease-specific survival; RT, radiotherapy.

  2. a

    Derived using the log-rank test of trend.

Entire cohort13080% 
Male85 (65)82%.46
Female45 (35)78% 
Extent of desmoplastic component   
Pure desmoplastic100 (77)78%.28
Mixed desmoplastic30 (23)86% 
T classification   
T11 (1)100%.26a
T215 (12)100% 
T332 (25)74% 
T469 (53)79% 
Lymph nodes at diagnosis   
No125 (96)80%<.001
Yes5 (4)<20% 
Head and neck80 (62)75%.05
Trunk/extremity50 (38)89% 
Presented with recurrent disease after prior surgery   
No110 (85)75%.32
Yes20 (15)90% 
Primary local management   
Surgery59 (45)84%.83
Surgery plus RT71 (55)77% 
Final resection margins   
Negative121 (93)80%.001
Positive/uncertain9 (7)44% 
Figure 1.

Disease-specific, disease-free, and overall survival rates are shown for the entire cohort.

Patterns of Disease Recurrence and Survival Implications

Of the 130 patients, 35 (27%) developed disease recurrence. Nineteen patients (15%) experienced LR. The actuarial LR rate was 17% at 5 years and beyond. The interval between surgery and LR ranged from 1 month to 60 months (median, 7 months); all instances of LR were evident by 5 years. Of all 19 patients who developed LR, it was the first manifestation of disease recurrence in 15 patients and 4 others developed LR after initial disease recurrence at a metastatic site. Nine of the 15 patients (60%) with apparently isolated LR developed subsequent distant metastases. Fifteen patients (12%) developed lymph node recurrence at a median time of 12 months (range, 1 month-8.5 years). The actuarial rates of lymph node recurrence at 5 years and 10 years were 11% and 14%, respectively. There was no significant difference in lymph node recurrence between patients with pure and mixed desmoplastic melanoma (12% and 11% respectively, at 5 years; P = 81). Twenty-seven patients (21%) developed distant metastases at a median of 19 months (range, 1.8 months-103 months). The actuarial rate of distant metastases development at 5 years and 10 years was 20% and 25%, respectively. Among the 27 patients who developed distant metastases, this was the site of first disease recurrence in 11 and the site of a subsequent disease recurrence in the remaining 16 patients. Patients who presented with involved lymph nodes at the time of diagnosis were at a higher risk of distant metastases than those who did not (P < .0001).

The median OS and DSS times after first disease recurrence were both 20 months. The actuarial DSS was only 33% at 3 years after recurrence. We could find no disease recurrence pattern that significantly correlated with DSS after recurrence. In particular, there was no difference in DSS after disease recurrence between those with ostensibly isolated LR and those with metastatic disease, most likely reflecting the high rate (60%) of subsequent distant metastases among patients who developed an LR. The median postrecurrence survivals for those with an initial LR and those with metastases were 29 months and 13 months, respectively (P = .23).

Local Control, Perineural Invasion, and the Role of RT

Of the 59 patients who underwent surgery without postoperative RT, 14 (24%) experienced LR. Of the 71 patients with were treated with surgery and postoperative RT, 5 (7%) experienced LR. Figure 2 shows local control by local treatment strategy. Four patients developed LR within the RT field, and 1 recurrence occurred at the margin of the RT field. LR as a manifestation of local spread along a nerve was clearly documented in only 1 patient and was noted among those patients treated with surgery alone. Table 2 shows the univariate analysis of factors potentially predictive of LR. In a Cox multivariate regression model, factors found to be significant for improved local control included the receipt of postoperative RT (P = .03) and negative resection margins (P = 0.008).

Table 2. Local Control and DMFS: Univariate Analysis for All 130 Patients
CharacteristicNo. (% of Total)10-Year LC RateP10-Year DMFS RateP
  1. Abbreviations: DMFS, distant metastasis-free survival; LC, local control; RT, radiotherapy.

  2. a

    Derived using the log-rank test of trend.

Entire cohort13083% 75% 
Male62 (52)86%.2475%.94
Female57 (48)78% 74% 
Extent of desmoplastic component     
Pure desmoplastic100 (77)82%.4573%.35
Mixed desmoplastic30 (23)87% 86% 
T classification     
T11 (1)100%.79a100%0.11a
T215 (12)100% 100% 
T332 (25)76% 72% 
T469 (53)85% 73% 
Head and neck80 (62)80%.1967%.01
Trunk/extremity50 (38)89% 87% 
Presented with recurrent disease after prior surgery     
No110 (85)85%.4974%.99
Yes20 (15)74% 80% 
Primary local management     
Surgery59 (45)74%.00978%.36
Surgery plus RT71 (55)91% 72% 
Final resection margins     
Negative121 (93)86%.00176%.19
Positive/uncertain9 (7)49% 61% 
Primary local management stratified by extent of desmoplastic component     
Pure desmoplastic     
Surgery46 (35)70%.00273%.83
Surgery plus RT54 (42)92% 72% 
Mixed histologic subtyping with desmoplastic features     
Surgery13 (10)88%.66100%.07
Surgery plus RT17 (13)88% 76% 
Figure 2.

Local control is shown by local treatment strategy, demonstrating that patients who received postoperative radiotherapy had superior local control (P = .009).

Local control was not found to be significantly different among those with perineural invasion compared with those without (81% vs 88%, respectively; P = .52). However, when local control outcomes were stratified by whether the patient had received postoperative RT, patients with perineural invasion who received postoperative RT had significantly better local control than those who did not (91% vs 63% at 10 years, respectively; P = .02). Among the 35 patients for whom it was clearly documented that there was no neurotropism associated with their melanoma, there was no significant difference noted with regard to local control when comparing those who received postoperative RT and those who did not (90% vs 85% at 10 years, respectively; P = .82). Only 5 patients were found to have neural invasion of a named nerve. Four of these patients received postoperative RT and 1 did not. Local control for these patients (81% at 10 years) was not significantly inferior compared with patients who did not have perineural invasion or who had perineural invasion but not of a named nerve (P = .69).

Complications of Treatment

Twenty-one patients (16%) experienced surgical complications. Of these, 11 were considered moderate in severity (6 patients with delayed wound healing or wound infection that required medical or outpatient procedural management, 1 patient with ear pain, 1 patient with edema, 1 patient with lagophthalmos, 1 patient with oral incompetence due to a surgical defect after lip surgery, and 1 patient who developed nosocomial pneumonia while recovering in the hospital postoperatively). Ten patients experienced surgical complications that were considered severe (3 patients with hematoma in the operative bed requiring evacuation and ligation under anesthesia, 2 patients with a surgical defect in the lip/microstomia requiring plastic surgery revision, 1 patient who experienced graft failure, 1 patient with nerve damage, 1 patient who experienced a cerebral vascular event while recovering in the hospital postoperatively, 1 patient with facial asymmetry and ectropion requiring reconstructive surgery, and 1 patient with redundant tissue in the scalp flap requiring plastic surgery revision).The actuarial rate of surgical complications was 16% at 5 years, and the median time to a surgical complication was 1 month (range, 0 months-16 months).

Of the 71 patients who received postoperative RT, 15 (21%) experienced a RT-related complication. The median time to the development of RT-related complication was 19 months (range, 1 month-12.5 years). Two patients had complications that were considered mild (1 patient with skin fibrosis and 1 patient with delayed wound healing). Seven patients had complications that were considered moderate (2 patients with hypothyroidism, 2 with delayed wound healing, 1 patient with edema, 1 patient with xerostomia, and 1 with keratoconjunctivits sicca/blepharatis). Six patients had severe RT-related complications (3 patients with osteoradionecrosis, 2 patients with nonhealing scalp wounds requiring surgical revision, and 1 patient with skin graft failure). The actuarial rates of significant RT-related complications (defined as moderate or severe) were 18% and 22%, respectively, at 5 years and 10 years.


In the current series, we observed that the overall rate of LR for desmoplastic melanoma was higher than that associated with nondesmoplastic cutaneous melanoma, which ranges from 1% to 5% after wide local excision alone.[16] However, when RT was delivered postoperatively, the LR rate decreased significantly. Other investigators have also reported, albeit to the best of our knowledge none in a randomized study, that RT improves local control for desmoplastic melanoma. Foote et al[10] reviewed their series of patients with desmoplastic melanoma and found a 3-year infield recurrence rate of only 9% with the use of adjuvant RT. Chen et al[11] suggested that RT may improve local control outcomes for desmoplastic melanoma by noting that patients in their series who were treated with RT who had more advanced tumors and smaller resection margins (ie, those with a high risk of LR) had local control rates similar to those of patients who had thinner tumors resected with wide margins. Similarly, Vongtama et al[4] reported an improvement in local control outcomes among patients who received adjuvant RT for either primary presentations or local recurrence of desmoplastic melanoma. However, occasional reports have challenged the need for adjuvant RT after wide local excision with adequate surgical margins. Specifically, Arora et al[9] reported an LR rate of only 4% in their recent series of 65 patients treated with wide excision and no postoperative RT.

With respect to which characteristics might portend poorer local control for patients with desmoplastic melanoma, some investigators have reported that the presence of neurotropism conferred a higher risk of LR.[5, 17] However, our finding that pathologically identified neurotropism was not adversely prognostic is similar to the findings of other more recent studies that did not find perineural invasion to be associated with a higher risk of LR.[4, 10, 18, 19] Our finding that a head and neck location was not associated with a higher risk of LR is similar to the finding by investigators at University of California at Los Angeles[4] in their cohort of 44 patients treated with surgery and RT for desmoplastic melanoma. However, Chen et al[11] did find that a head and neck location was associated with a significantly higher risk of LR among their cohort of 128 patients with desmoplastic melanoma, and although not all of the patients in their analysis received RT, 96% of those with head and neck primary tumors did undergo adjuvant irradiation, suggesting that their finding held even in the setting of aggressive local therapy.

In 2004, Busam et al and Hawkins et al published their reports that introduced the concept that the extent of desmoplastic component in a melanoma lesion carried prognostic significance.[12, 13] Specifically, they posited that patients with pure desmoplastic melanoma, those tumors in which the desmoplasia is extensive throughout the tumor, have a better DSS than patients in whom the tumors demonstrate a desmoplastic component that is present only in a percentage of an otherwise nondesmoplastic melanoma (ie, combined/mixed desmoplastic melanoma). Murali et al[2] specifically explored this issue in their study of 252 patients with desmoplastic melanoma in which they reported that whether the tumor was pure versus combined desmoplastic melanoma was not found to be predictive of DFS but that patients with combined desmoplastic melanomas had a higher risk of locoregional recurrence and shorter times to disease recurrence. Maurichi et al[7] reviewed 242 cases of desmoplastic melanoma with a specific focus on the prognostic implications of the extent of desmoplastic component (pure vs mixed). They demonstrated that patients with mixed desmoplastic melanomas experienced higher rates of regional and distant recurrences, with outcome patterns similar to those of patients with nondesmoplastic melanoma, whereas patients with pure desmoplastic melanoma had significantly lower rates of regional or distant metastases but a significantly higher risk of LR if local therapy was not aggressive (eg, resection margins were inadequate). In the current study, we did not find any difference with regard to DSS, distant metastases-free survival, or local control with respect to the extent of desmoplasia in the tumor. However, it is interesting to note that among the subgroup of patients with pure desmoplastic melanoma, the addition of adjuvant irradiation to local therapy significantly improved local control. Further study is needed to determine whether there is a threshold percentage of desmoplasia above which RT may be particularly significant in improving local control.

There are limitations to the interpretation of the findings from the current retrospective cohort study. Specifically, the number of patients with combined/mixed desmoplastic melanoma in this study was small, which limited our ability to draw firm conclusions regarding the influence of the extent of desmoplasia on outcomes. Because the local treatment approach was at the discretion of treating surgeons and radiation oncologists over a study window spanning multiple treatment eras, selection bias inherently played a role in determining which patients received RT versus who did not. Furthermore, the current study included patients assessed and treated in the era before the extent of desmoplastic component as a percentage of the specimen was routinely quantified, thus limiting our ability to specifically inform on what extent of desmoplasia may prognosticate outcomes. However, to the best of our knowledge, all the existing studies of this relatively rare malignancy are retrospective cohort studies. Currently, the Trans Tasman Radiation Oncology Group has undertaken an important randomized trial examining the role of postoperative RT for patients with neurotropic melanomas of the head and neck (TROG 08.09) that will contribute to our understanding of its role as an adjuvant therapy for desmoplastic melanoma because neurotropism is commonly observed in these patients.[20] However, until such randomized data are available, we continue to rely on the published clinical cohort studies for this entity.

The optimal dosing regimen for postoperative RT has yet to be determined, and conventional dosing for postoperative RT of 60 Gy in 30 fractions is an acceptable radiotherapeutic approach. The recently published TROG 02.01 trial that demonstrated improved regional control for patients receiving 48 Gy in 20 fractions postoperatively for high-risk regional lymph node melanoma metastases also offers a well-tested dosing regimen for patients with indications for the postoperative treatment of melanoma.[21] Accepted regimens also include the regimen used in the current study of 30 Gy in 5 fractions delivered twice weekly over 2.5 weeks, usually using appositional electrons for this cutaneous, and therefore superficial, malignancy. On a technical note, if this regimen is used, the dosing of electrons is prescribed to D-maximum (point dose).[22] Overall treatment-related complications may be relatively increased by the addition of postoperative RT, especially given that desmoplastic melanoma frequently occurs in the head and neck, which presents anatomical challenges for aggressive tumor eradiation. To minimize unnecessary complications, RT volumes should be configured with an understanding that this is a cutaneous malignancy and large treatment volumes aimed at covering the paths of nerves are not warranted. Data from the current study as well as those of others suggest that successful local therapy is critical to optimizing overall outcomes for patients with desmoplastic melanoma.

The results of the current study indicated that combined modality local therapy that included wide local excision followed by RT provided superior local therapy compared with surgery alone for patients with desmoplastic melanoma. Furthermore, negative resection margins are critical for cancer control outcomes (both local control and DSS), even when postoperative RT is administered. Complications may be increased through the use of aggressive combined modality therapy, but this must be weighed against the consideration that uncontrolled local recurrence can be morbid or portend a higher risk of subsequent disease dissemination. Further study is needed to evaluate the role of the extent of desmoplasia and control outcomes with respect to the use of adjuvant RT.


No specific funding was disclosed.


Dr. Ross is a member of the board of Merck Inc and has received fees as a member of the Speakers' Bureau for Merck, Amgen, and Genomic Health.