July 2017
Volume 58, Issue 9
Open Access
Anatomy and Pathology/Oncology  |   July 2017
Comparison of Alternative Tumor Size Classifications for Posterior Uveal Melanomas
Author Affiliations & Notes
  • Cassandra C. Skinner
    Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States
  • James J. Augsburger
    Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States
  • Bret D. Augsburger
    Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States
  • Zelia M. Correa
    Department of Ophthalmology, University of Cincinnati College of Medicine, Cincinnati, Ohio, United States
  • Correspondence: Zelia M. Correa, Department of Ophthalmology, Medical Science Building 5th Floor, 231 Albert Sabin Way, Cincinnati, OH 45267–0527, USA; correazm@uc.edu
Investigative Ophthalmology & Visual Science July 2017, Vol.58, 3335-3342. doi:10.1167/iovs.16-20465
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Cassandra C. Skinner, James J. Augsburger, Bret D. Augsburger, Zelia M. Correa; Comparison of Alternative Tumor Size Classifications for Posterior Uveal Melanomas. Invest. Ophthalmol. Vis. Sci. 2017;58(9):3335-3342. doi: 10.1167/iovs.16-20465.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Determine which posterior uveal melanoma (PUM) size classification with three categories has the best prognostic discrimination.

Methods: Single-institution study of 424 consecutive patients with PUM. The tumor's largest basal diameter (LBD), smallest basal diameter (SBD), and thickness (TH) were estimated by fundus mapping and ultrasonography. Tumors were assigned to “small,” “medium,” or “large” size categories defined by 11 different classifications (Linear LBD, Rectangular LBD × TH, Cubic LBD × SBD × TH, Warren Original, Warren Modified, Augsburger, COMS Original, COMS Revised, TNM 2002, and modified TNM 2010 classification [a,b]). Prognostic significance of classifications was evaluated by Kaplan-Meier event curves with computation of log rank test for trend statistic.

Results: In six classification systems (Warren Original, Warren Modified, COMS Revised, TNM 2002, TNM 2010a, TNM 2010b) >50% of tumors fell within one subgroup. In the Warren Original classification <5% of tumors fell within one subgroup. Separation of Kaplan-Meier curves among three size categories was judged “excellent” in four classifications (Linear LBD, Cubic Volume, TNM 2010a, and TNM 2010b) and “very poor” in the Warren Original. Linear LBD classification was associated with highest log rank statistic value. TNM 2010a, TNM 2010b, TNM 2002, Augsburger, and Cubic Volume classifications were also determined to be quite good.

Conclusions: Linear LBD classification was the best three-size category discriminator among low-, intermediate-, and high-risk subgroups. Considering our findings, it seems possible that the arduous work required to apply complex classifications, especially for three-category systems, for PUM may not be justified in routine clinical practice.

Tumor size classification of malignant neoplasms is used clinically as a therapeutic and prognostic tool.13 Size classification determines a patient's eligibility for a specific form of treatment or clinical trial. It can also be used to assign patients to prognostic subgroups, thus allowing physicians to engage in reasonably accurate discussions with patients about the actuarial probability of developing metastasis impacting their survival. 
Numerous size-based classification systems have been developed for posterior uveal melanoma (PUM). Some classifications use largest basal diameter (LBD) of the tumor alone,47 many take into account both LBD and tumor thickness (TH),819 and at least one takes into account smallest basal diameter (SBD) as well as LBD and TH.2022 
This study was designed to evaluate multiple reported and some derived tumor size classifications of PUM that separate tumors into small, medium, and large categories and determine which three-size category classification provides the best prediction of low, intermediate, and high risk for metastasis. 
Methods
Our study population consisted of 424 consecutive patients with PUM (choroidal, ciliochoroidal, or iridociliochoroidal) who were evaluated and managed at a single center from 2007 through 2014. All patients were evaluated by fine-needle aspiration biopsy and had confirmed cytologic (or histopathologic) diagnosis of PUM. Tumor dimensions (LBD, SBD, and TH) were estimated in all cases by cartographic fundus mapping and ocular ultrasonography. Baseline clinical variables evaluated in this study included patient age and sex, and tumor dimensions and location (choroidal, ciliochoroidal, or iridociliochoroidal). 
Review of the published literature identified eight tumor size classifications of PUM that separated tumors into small (lowest probability of metastasis expected), medium (intermediate probability of metastasis expected), and large (highest probability of metastasis expected) categories (Linear LBD, Cubic Volume, Warren Original classification, Warren Modified classification, Augsburger classification, Collaborative Ocular Melanoma Study [COMS] Original classification, COMS Revised classification, and TNM 2002 classification).46,813,1517,20,23 The authors formed a three-category Rectangular LBD × TH classification by combining two “two-category” LBD × TH classifications (one defining a “small” tumor as LBD × TH ≤30 mm2 and regarding all others as large,24 another defining a “large” tumor as LBD × TH >100 mm2 and regarding all others as small).19 Finally, two additional three-category classifications were created by merging subgroups of the four-category TNM 2010 classification25 as follows: TNM 2010a combined T1 and T2 subgroups, and TNM 2010b combined T2 and T3 subgroups. 
Tumor classification graphs demonstrating the boundary lines for the separation of small, medium, and large tumors were prepared for each of the 11 classifications (Fig. 1). Tumors were then plotted on each classification's boundary lines and assigned a tumor size subgroup (i.e., small, medium, or large) according to each system. Tumors falling on a boundary line were assigned arbitrarily to the lower risk group. 
The Ease Versus Difficulty of Applying the Classification to a Group of Tumors
The ease versus difficulty of applying a size classification to a group of tumors was rated from easiest (0 points) to most difficult (2 points) based on whether mathematical manipulation of the raw data and/or reference to a graph was required. If mathematical manipulation of raw data was not required, 0 points were given. If mathematical manipulation (e.g., multiplication of LBD × TH) was required, then 1 point was given. If reference to a graph of intergroup boundary lines was not required, 0 points were given. If reference to category boundary lines was required to determine the size category, then 1 point was given. 
Conventional descriptive statistics (mean, standard deviation of mean, minimum, maximum) were computed for all the evaluated continuous numerical variables (LBD, SBD, TH, rectangular area [LBD × TH] of tumor, cubic volume [LBD × SBD × TH] of tumor, age of patient) and frequency distributions for all study variables, including categorical variables. 
The principal outcome evaluated was development of metastasis during follow-up. No patient had evidence of metastasis at the time of initial treatment. Data analysis extended to December 31, 2015. 
Kaplan-Meier event rate curves for metastasis according to tumor size category assigned by each size classifications were computed and plotted. The effectiveness of each classification for assigning patients to discreet ordered low-, medium-, and high-risk categories for metastasis was assessed by (1) inspection of the plotted event rate curves to determine how clearly separated the three curves were and whether the curves were ordered as expected (i.e., showing lowest risk for the cases categorized as small, highest risk for the cases categorized as large, and intermediate for the cases categorized as medium), and (2) computing a log rank test of trend statistic for the significance of the separation of the curves plotted for each classification. The null hypothesis was that the curves were substantially different from one another and ordered as expected. The greater the separation of the ordered event rate curves, the higher the value of the computed test statistic. 
Results
Baseline Patient and Tumor Characteristics
Features of the 424 patients and their PUM are summarized in Table 1. Seventy-two percent of studied tumors were choroidal (n = 306), 23.8% were ciliochoroidal (n = 101), and 4% were iridociliochoroidal (n = 17). 
Table 1
 
Tumor and Population Descriptive Statistics.
Table 1
 
Tumor and Population Descriptive Statistics.
Frequency Distribution of Tumors According to Classification
The frequency distributions of tumors assigned to size categories by the 11 evaluated classifications are summarized in Table 2. Six classification systems (Warren Original, Warren Modified, COMS Revised, TNM 2002, TNM 2010a, TNM 2010b) had more than 50% of tumors falling within one subgroup and another classification system, the Warren Original, had fewer than 5% of tumors falling within one subgroup. 
Table 2
 
Definitions of Tumor Size Categories by Each Evaluated Tumor Size Classification and Frequency Distribution of Small, Medium, and Large Tumors Assigned by Each Classification to 424 PUMs
Table 2
 
Definitions of Tumor Size Categories by Each Evaluated Tumor Size Classification and Frequency Distribution of Small, Medium, and Large Tumors Assigned by Each Classification to 424 PUMs
Tumor Size Category of the “Average Tumor” In the Group
The “average tumor” in our study (i.e., hypothetical tumor having the mean value of LBD, SBD, and TH of all tumors in this series [12.2 × 10.5 × 5.8 mm]) was categorized as “small” by one classification (TNM 2010a), “medium” by eight classifications (Linear LBD, Rectangular Area, Cubic Volume, Augsburger, COMS Original, COMS Revised, TNM 2002, and TNM 2010b), and “large” by two classifications (Warren Original, Warren Modified). 
Separation of Event Curves
Kaplan-Meier curves for the three tumor size subgroups by each evaluated classification are shown in Figure 2. Inspection of these curves shows considerable differences in discrimination between small, medium, and large tumors based on each individual classification. The separation between curves was rated as indicated in Table 3. Linear LBD, Cubic Volume, TNM 2010a, and TNM 2010b classifications were all rated “excellent,” whereas the Warren Original classification was rated “very poor.” 
Figure 1
 
Boundary line graphs for separation of PUMs into small, medium, and large categories for the tumor size classifications evaluated in this study. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Figure 1
 
Boundary line graphs for separation of PUMs into small, medium, and large categories for the tumor size classifications evaluated in this study. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Figure 2
 
Kaplan-Meier graphs of cumulative probability of uveal melanoma metastasis for patients with tumors assigned to small, medium, and large categories by the evaluated tumor size classifications. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Figure 2
 
Kaplan-Meier graphs of cumulative probability of uveal melanoma metastasis for patients with tumors assigned to small, medium, and large categories by the evaluated tumor size classifications. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Table 3
 
Comparison of Tumor Size Classifications of PUMs Evaluated in This Study
Table 3
 
Comparison of Tumor Size Classifications of PUMs Evaluated in This Study
Ease of Classification Application
A scoring system from 0 to 2 is listed in Table 3 for the evaluated classification systems. 
Prognostic Significance of Evaluated Classification Systems
Table 4 lists the evaluated classifications in rank order by the log rank test of trend statistic. The highest the log rank test statistic was associated with the Linear LBD classification, whereas the lowest log rank test statistic was associated with the Rectangular LBD × TH classification. 
Table 4
 
Ranking of Tumor Size Classifications by Log Rank Test of Trend Statistic From Kaplan-Meier Event Rate Curves Applied to 424 Cases of PUM
Table 4
 
Ranking of Tumor Size Classifications by Log Rank Test of Trend Statistic From Kaplan-Meier Event Rate Curves Applied to 424 Cases of PUM
Discussion
It is surprising that tumor size was not recognized to be a prognostic factor for metastasis and metastatic death in PUM until approximately the middle of the 20th century.22 Classic articles by Callender and colleagues26,27 in the 1930s through 1940s describe histopathologic features of PUM in enucleated eyes but do not mention tumor size. The earliest article mentioning tumor size as a prognostic factor that found in our literature search divided the PUMs into “smaller” and “larger” subgroups at the mean of the cubic tumor volume of the evaluated tumors in a series of enucleated eyes (LBD2 × TH = 1344 mm3).22 The authors indicated that metastasis and metastatic death were substantially higher in the subgroup with larger tumors. 
The first three-category (i.e., small, medium, and large) size classification of PUM was developed in the 1950s.28 This classification, the Linear LBD, has been used in multiple publications since that time.46 This classification has only one size parameter, making it the easiest among the tumor size classifications evaluated in this study to use. Although we could find no explanation why LBD of 10 and 15 mm were selected as boundary lines for this classification, our study shows this classification system to be the best prognostic indicator of metastasis-free survival by the log rank test of trend. 
In 1963, Jensen29,30 categorized PUM as large if LBD × TH was >100 mm2 and small if LBD × TH was ≤100 mm2. He continued to use this two-category classification through the 1980s showing that patients with larger tumors had worse survival prognosis than those with small tumors.19 
In 1975, Davidorf and Lang24 defined small PUM as having LBD × TH ≤30 mm2 and all other tumors as large. In 1979, Thomas and colleagues21 used this same definition for small versus large PUMs. Both groups reported that patients having a small PUM had an extremely favorable prognosis for metastasis-free survival. As previously mentioned, these two “two-category” classifications were combined to form an evaluable three-category “Rectangular Area” classification with boundary lines at 30 mm2 and 100 mm2. By using this system, tumors within our cohort were fairly well distributed among the subgroups. Unfortunately, despite the ease of using this system, the prognostic significance was relatively low compared with the other systems we evaluated. 
In 1980, Gass20 devised a classification system using tumor cubic volume (LBD2 multiplied by TH). Without a written rationale, he categorized small tumors as ≤300 mm3, medium as >300 mm3 but ≤1125 mm3, and large as >1125 mm3. His “Cubic Volume” method adequately distributed the cohort among the subgroups and provided satisfactory discrimination among low-, intermediate-, and high-risk cases. 
In 1974, Warren8 published his original classification system for PUM that combined LBD and TH. In his classification, tumors were categorized as small if LBD was ≤10 mm and TH was ≤2 mm, large if LBD was >15 mm or TH was >5 mm, and medium if dimensions were between those two extreme categories. Warren8 justified his selection of his boundary lines by treatment guidelines used at the time. His system is relatively simple to use, having only two parameters and a relatively straightforward graphical representation. One drawback of the original Warren classification is the somewhat restrictive small and medium criteria categorizing most tumors as large. Furthermore, this classification does not provide strong prognostic significance based on subgroups generated. 
By the mid to late 1970s, several ophthalmologists were using a modified Warren classification in which the thickness boundary line between small and medium tumors was changed from 2 mm to 3 mm.912 Individual authorship and rationale for this modification has not been established in the peer-reviewed literature, making it impossible to determine why this change was suggested. Like the original Warren classification, this modified system still assigned the average tumor to the large category principally based on tumor thickness. 
In 1993, Augsburger13 proposed an alternative system to the modified Warren classification attempting to decrease the impact of tumor TH in the categorization of PUM. He observed on a scatterplot of TH by LBD that PUM, TH of a typical PUM was approximately half the LBD. This finding influenced the change in boundaries used by the Warren classification but chose 5.0 and 7.5 mm as TH boundaries. He retained 10 and 15 mm as LBD boundary lines. Although this classification succeeded in creating nearly equivalent patient subgroups, its prognostic significance was not as good as that of the Linear LBD system. 
In 1988, the COMS planning group devised a tumor size classification used to direct patients into different treatment arms of that study.14 The original boundary lines for the small, medium, and large categories in this study are shown in Figure 1 (COMS original classification). In 1990 the COMS group modified this classification by moving the tumor thickness boundaries from 3 and 8 mm to 2.5 and 10 mm (Fig. 1, COMS Revised classification).23 Both classifications used LBD of 16 mm as the dividing boundary between large and smaller tumors. Additionally, both COMS classifications defined nevi as melanocytic choroidal tumors ≤5 mm in LBD and ≤1 mm in TH. Bear in mind that the COMS classifications were not developed for prognostic purposes but rather for therapeutic decision-making. In spite of this, these classifications perform reasonably well in discriminating patients as low, medium, and high risk of metastasis. Both classifications place most patients in the medium size category. 
In 2002, the American Joint Committee on Cancer (AJCC) released a new staging system for PUM.31,32 This staging system took into account the size of the primary tumor (T), presence of lymph node involvement (N), presence of distant metastasis (M), and the presence of extrascleral tumor extension. The 2002 TNM system for malignant uveal melanoma defined small (T1) tumors as ≤10 mm in LBD and ≤2.5 mm in TH, large (T3) tumors as >16 mm in LBD or >10 mm in TH, and medium (T2) tumors as intermediate in size. This system categorized the “average tumor” into the medium (T2) subgroup but assigned only 10.6% of our studied patients to the small (T1) category. The prognostic strength of the TNM 2002 was moderate when compared with other systems evaluated. 
In 2010, the AJCC released the seventh edition of its cancer staging manual.25 In this revision, tumor size was subdivided into four rather than three size categories (see Fig. 1, TNM 2010 classification). The boundary lines for LBD and TH for the four T categories were derived empirically from a collaborative database of more than 7000 patients with uveal melanoma. Because our study evaluated only classifications that divided patients into three size subgroups, two modifications of the TNM 2010 classification were generated by combining specified TNM 2010 tumor size categories. Although the frequency distribution of tumors was skewed to the small category for TNM 2010a and to the medium category for TNM 2010b, both systems exhibited excellent discrimination between the event curves and reasonably good prognostic significance compared with most other systems we evaluated. 
When evaluating and comparing tumor size classifications of PUM, several features of each classification must be taken into account, including those described in the following sections. 
The Size Category (Small, Medium, or Large) to Which the Classification Assigns an “Average Tumor”
As long as the patient group and tumors being evaluated in a study are representative of the entire size spectrum of PUM and not just a limited tumor size subgroup, a satisfactory classification should always assign the “average tumor” (i.e., hypothetical tumor having the mean LBD, SBD, and TH for the entire group by that classification) to the medium size category. It is important to point out that several of the classifications evaluated (Table 3) did not achieve this goal. 
The Proportion of Tumors Assigned to Small, Medium, and Large Categories by the Classification
In our opinion, a satisfactory classification of PUM should assign nearly equivalent numbers of patients to each of the tumor size categories. In this study, we regarded a tumor size classification as satisfactory if no subgroup contained <20% of the cases, or >50% of the cases. Six of the evaluated classifications were judged very unsatisfactory (Table 3) by this aspect of its separation of cases in this study. 
The Ordered Separation Between Event Rate Curves of the Patients Assigned to the Small, Medium, and Large Categories by the Classification
A satisfactory tumor size classification should separate tumors into subgroups that yield distinct low, intermediate, and high-risk event rate curves. A classification that provides overlapping curves for any of the tumor size subgroups (Warren classifications, COM Revised classification) or a lower event curve for a larger size subgroup (original Warren classification) than for a smaller size subgroup cannot be regarded as an effective classification. 
The Ease of Application of the Classification System
A linear boundary line classification separating tumors based on a single dimension (e.g., Linear LBD classification) is clearly easiest to apply. Such a classification does not necessitate reference to a graph or mathematical manipulation of the original data. Nonoverlapping “single-step” boundary line classifications (e.g., Warren Original, Warren Modified, Augsburger, TNM 2002) that subdivide tumors on the basis of both LBD and TH with discrete nonoverlapping small-medium and medium-large boundary lines are more difficult to apply given the need to reference a graph. Nonoverlapping “multiple-step” classifications (e.g., TNM 2010a and TNM 2010b) and classifications that have overlapping boundary lines (e.g., COMS Original and COMS Revised) that subdivide tumors on the basis of both LBD and TH are also more difficult to apply given the need for graph reference. These classifications require plotting of each tumor individually on predetermined category boundary lines. Curvilinear boundary line classifications separating tumors based on the product of two tumor dimensions (rectangular area classification, LBD × TH) or three tumor dimensions (cubic volume classification, LBD × SBD × TH) are most difficult to apply given the required mathematical manipulation followed by reference to a graph. In these cases, the original data need to be computed before these values can be used to separate the cases. In general, an easy-to-apply classification that also provides good proportional separation of the cases and good prognostic discrimination of the event curves for those subgroups is more effective in clinical situations. 
The Prognostic Value of the Classification With Regard to the Event Rates of Interest
The clinical significance of a tumor size classification (i.e., its prognostic value) takes into account the magnitude of the separation among the relevant event curves for the tumor size subgroups as well as the proportional sizes of the subgroups. This significance was evaluated in this study using the logrank test of trend. This statistical method evaluates both the separation of the curves and whether they are ordered as expected. The larger the test statistic provided by the classification, the stronger the classification as a prognostic variable (Table 4). 
Caveats
In our application of the TNM 2010 tumor size classification, we combined tumor size subgroups in ways that were not intended by the individuals who developed this classification. Readers should understand that our purpose was to make the TNM 2010a and TNM 2010b classifications directly comparable to the other classifications we evaluated. The subclassifications developed may not fully encompass the prognostic ability of this staging system compared with other classifications evaluated. Similarly, the Rectangular Area classification evaluated in this study was devised by different authors.19,24 The new three-part classification we generated may not reflect the original intent of the authors. We also recognize that the results of a study such as ours could possibly differ if applied to other PUM populations. 
In conclusion, our study does not “anoint” any individual tumor size classification as the “optimal” or “best” system or provide sufficient evidence to replace all other classifications by a single system. However, it points out the limitations of many classifications and provides sufficient evidence to suggest that some of them are insufficiently discriminating to be in use. Although some classifications contain the best combinations of favorable ratings for each evaluated factor, overall the Linear LBD classification had the highest predictive value for PUM metastasis-free survival. Considering the Linear LBD system's ease of application and prognostic value determined by this study of three-category classification systems, it seems debatable whether more intricate classification systems are really necessary in routine clinical practice. 
Acknowledgments
Supported in part by an unrestricted grant from Research to Prevent Blindness, Inc., New York, NY, USA, to the Department of Ophthalmology, University of Cincinnati College of Medicine; the Augsburger Ocular Oncology Fund of the University of Cincinnati College of Medicine; the Dr. E. Vernon and Eloise C. Smith Chair of Ophthalmology Fund of the University of Cincinnati; and the Dr. Mary Knight Asbury Chair of Ophthalmic Pathology and Ocular Oncology Fund of the University of Cincinnati. 
Disclosure: C.C. Skinner, None; J.J. Augsburger, None; B.D. Augsburger, None; Z.M. Correa, Castle Biosciences, LLC (C, R) 
References
Fritz AG. International Classification of Diseases for Oncology: ICD-O. 3rd ed. Geneva: World Health Organization; 2000.
Marchegiani G, Andrianello S, Malleo G, et al. Does size matter in pancreatic cancer? Reappraisal of tumour dimension as a predictor of outcome beyond the TNM. Ann Surg. 2016; 18 (suppl 1): e347.
Saito H, Osaki T, Murakami D, et al. Macroscopic tumor size as a simple prognostic indicator in patients with gastric cancer. Am J Surg. 2006; 192: 296–300.
McLean IW, Foster WD, Zimmerman LE. Uveal melanoma: location, size, cell type, and enucleation as risk factors in metastasis. Hum Pathol. 1982; 13: 123–132.
Packard RB. In malignant choroidal melanoma will a delay in radical treatment influence prognosis? Trans Ophthalmol Soc U K. 1983; 103: 49–53.
Coleman K, Baak JP, Van Diest P, Mullaney J, Farrell M, Fenton M. Prognostic factors following enucleation of 111 uveal melanomas. Br J Ophthalmol. 1993; 77: 688–692.
Taktak AF, Fisher AC, Damato BE. Modelling survival after treatment of intraocular melanoma using artificial neural networks and Bayes theorem. Phys Med Biol. 2004; 49: 87–98.
Warren RW. Prognosis of malignant melanomas of the choroid and ciliary body. In: Blodi FH, ed. Current Concepts in Ophthalmology. St. Louis: CV Mosby; 1974: 158–166.
Shields JA. Diagnosis and Management of Intraocular Tumors. St. Louis: CV Mosby; 1983: 211.
Char DH. Clinical Ocular Oncology. New York: Churchill Livingstone; 1989: 87.
Gragoudas ES, Seddon JM, Egan K, et al. Long-term results of proton beam irradiated uveal melanomas. Ophthalmology. 1987; 94: 349–353.
Lommatzsch PK. Results after beta-irradiation (106Ru/106Rh) of choroidal melanomas: 20 years' experience. Br J Ophthalmol. 1986; 70: 844–851.
Augsburger JJ. Size classification of posterior uveal malignant melanomas. In: Liabson PR, ed. Year Book of Ophthalmology. St. Louis: CV Mosby; 1993: 155–165.
Straatsma BR, Fine SL, Earle JD, Hawkins BS, Diener-West M, McLaughlin JA. Enucleation versus plaque irradiation for choroidal melanoma. Ophthalmology. 1988; 95: 1000–1004.
The Collaborative Ocular Melanoma Study (COMS). Mortality in patients with small choroidal melanoma. COMS report no. 4. Arch Ophthalmol. 1997; 115: 886–893.
Diener-West M, Earle JD, Fine SL, et al. The COMS randomized trial of iodine 125 brachytherapy for choroidal melanoma, III: initial mortality findings. COMS Report No. 18. Arch Ophthalmol. 2001; 119: 969–982.
The Collaborative Ocular Melanoma Study (COMS). Randomized trial of pre-enucleation radiation of large choroidal melanoma II: initial mortality findings. COMS report no. 10. Am J Ophthalmol. 1998; 125: 779–796.
Barr CC, Sipperley JO, Nicholson DH. Small melanomas of the choroid. Arch Ophthalmol. 1978; 96: 1580–1582.
Jensen OA. Malignant melanomas of the human uvea: 25-year follow-up of cases in Denmark, 1943–1952. Acta Ophthalmol. 1982; 60: 161–182.
Gass JD. Observation of suspected choroidal and ciliary body melanomas for evidence of growth prior to enucleation. Ophthalmology. 1980; 87: 523–528.
Thomas JV, Green WR, Maumenee AE. Small choroidal melanomas. A long-term follow-up study. Arch Ophthalmol. 1979; 97: 861–864.
Flocks M, Gerende JH, Zimmerman LE. The size and shape of malignant melanomas of the choroid and ciliary body in relation to prognosis and histologic characteristics; a statistical study of 210 tumors. Trans Am Acad Ophthalmol Otolaryngol. 1955; 59: 740–758.
Accuracy of diagnosis of choroidal melanomas in the Collaborative Ocular Melanoma Study. COMS report no. 1. Arch Ophthalmol. 1990; 108: 1268–1273.
Davidorf FH, Lang JR. The natural history of malignant melanoma of the choroid: small vs large tumors. Trans Sect Ophthalmol Am Acad Ophthalmol Otolaryngol. 1975; 79: OP310–OP320.
Edge SB, Byrd DR, Comptom CC, Fritz AG, Greene FL, Troti A, eds. AJCC Cancer Staging Manual. New York: Springer; 2010.
Callender GR. Malignant melanotic tumors of the eye: a study histologic types in 111 cases. Trans Am Acad Ophthalmol Otolaryngol. 1931; 36: 131–142.
Callender GR, Wilder HC, Ash JE. Five hundred melanomas of the choroid body followed five years or longer. Am J Ophthalmol. 1942; 25: 962–967.
Flocks M, Gerende JH, Zimmerman LE. The size and shape of malignant melanomas of the choroid and ciliary body in relation to prognosis and histologic characteristics; a statistical study of 210 tumors. Trans Am Acad Ophthalmol Otolaryngol. 1955; 59: 740–758.
Jensen OA. Malignant melanomas of the human uvea: recent follow-up of cases in Denmark, 1943–1952. Acta Ophthalmol. 1970; 48: 1113–1128.
Jensen OA. Malignant melanoma of the uvea [in Danish]. Nord Med. 1963; 69: 502–503.
Kim CJ, Reintgen DS, Balch CM. The new melanoma staging system. Cancer Control. 2002; 9: 9–15.
Kujala E, Damato B, Coupland SE, et al. Staging of ciliary body and choroidal melanomas based on anatomic extent. J Clin Oncol. 2013; 31: 2825–2831.
Figure 1
 
Boundary line graphs for separation of PUMs into small, medium, and large categories for the tumor size classifications evaluated in this study. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Figure 1
 
Boundary line graphs for separation of PUMs into small, medium, and large categories for the tumor size classifications evaluated in this study. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Figure 2
 
Kaplan-Meier graphs of cumulative probability of uveal melanoma metastasis for patients with tumors assigned to small, medium, and large categories by the evaluated tumor size classifications. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Figure 2
 
Kaplan-Meier graphs of cumulative probability of uveal melanoma metastasis for patients with tumors assigned to small, medium, and large categories by the evaluated tumor size classifications. TNM 2010 classification is included to facilitate comparison of subclassifications generated (TNM 2010a and TNM 2010b).
Table 1
 
Tumor and Population Descriptive Statistics.
Table 1
 
Tumor and Population Descriptive Statistics.
Table 2
 
Definitions of Tumor Size Categories by Each Evaluated Tumor Size Classification and Frequency Distribution of Small, Medium, and Large Tumors Assigned by Each Classification to 424 PUMs
Table 2
 
Definitions of Tumor Size Categories by Each Evaluated Tumor Size Classification and Frequency Distribution of Small, Medium, and Large Tumors Assigned by Each Classification to 424 PUMs
Table 3
 
Comparison of Tumor Size Classifications of PUMs Evaluated in This Study
Table 3
 
Comparison of Tumor Size Classifications of PUMs Evaluated in This Study
Table 4
 
Ranking of Tumor Size Classifications by Log Rank Test of Trend Statistic From Kaplan-Meier Event Rate Curves Applied to 424 Cases of PUM
Table 4
 
Ranking of Tumor Size Classifications by Log Rank Test of Trend Statistic From Kaplan-Meier Event Rate Curves Applied to 424 Cases of PUM
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×