New Imaging Method Can Diagnose and Direct Treatment of Rare Pediatric Retinoblastoma

Ira Dunkel

Decreasing values for apparent diffusion coefficient (ADC) mapping derived from MRI may indicate the development of trilateral retinoblastoma in pediatric patients with bilateral retinoblastoma, according to our retrospective study published recently in the Journal of Pediatric Hematology/Oncology.

Based on our findings, we recommend that ADC values under 750 should prompt surgical consideration for trilateral retinoblastoma. Unchanged high values over 1,000 may distinguish those with benign pineal cysts, obviating invasive neurosurgical procedures. (1)

We compared ADC measurements from a group of six patients with proven trilateral retinoblastoma to those from a group of eight patients with bilateral retinoblastoma treated at Memorial Sloan Kettering Cancer Center. The median observed ADC for trilateral retinoblastoma was 620.95, compared with 1,238.5 for a normal pineal gland in bilateral retinoblastoma. (1)

Our preliminary investigation provides baseline reference data and illustrates the potential role for ADC mapping. We also reviewed three case studies where ADC mapping provided correct guidance on initially equivocal results. Overall, our report highlights the need for a prospective, multicenter study to validate the diagnostic accuracy of ADC trends in predicting the risk of trilateral retinoblastoma.

…we recommend that ADC values under 750 should prompt surgical consideration for trilateral retinoblastoma. Unchanged high values over 1,000 may distinguish those with benign pineal cysts, obviating invasive neurosurgical procedures.

Pediatric Retinoblastoma

Trilateral retinoblastoma is a well-defined yet rare pediatric cancer with an incidence of 3.5 percent in patients with the heritable form of retinoblastoma. (2) The term “trilateral retinoblastoma” refers to the association of bilateral retinoblastoma with an intracranial midline neuroblastic tumor. (3), (4) The incidence of pineal trilateral retinoblastoma is 4.2 percent in bilateral cases, and the odds of a nonpineal trilateral retinoblastoma are 0.8 percent. (2) The median age of diagnosis is 23 to 40 months. (3), (4), (5) For 20 to 25 percent of cases, the tumors are suprasellar or parasellar, and may exist before the diagnosis of retinoblastoma. (6)) The diagnosis of a pineal tumor usually occurs 20 months after the diagnosis of bilateral retinoblastoma, on average. (5), (6)

Historically, the prognosis for trilateral retinoblastoma has been poor; most patients die of their disease in less than nine months. (5), (7) Current treatment approaches, including intensive chemotherapy followed by myeloablative chemotherapy and autologous hematopoietic stem cell transplant, have improved survival outcomes but are associated with significant short-term and long-term toxicities. (8), (9)

Advanced Imaging Techniques

Patients are typically screened with MRI at baseline and as often as every six months for five years after diagnosis, with the hope of identifying trilateral retinoblastoma at a subclinical stage. (10) However, conventional MRI does not sufficiently capture the detail required to make a definitive diagnosis, especially as the goal is to identify small tumors as early as possible. Benign pineal cysts, which have a high incidence in children, regardless of retinoblastoma diagnosis, can confound test results. (11), (12)

ADC mapping is a newer imaging technique. It takes advantage of the biology of pineoblastomas, which have abnormally restricted water diffusion and show a high signal intensity on diffusion-weighted imaging (DWI). (13)

Study Findings

We examined serial MRI brain scans for 14 patients treated at MSK between July 1993 and March 2016 and compared ADC measurements from a group of six patients with biopsy-proven trilateral retinoblastoma to those from a group of eight patients with bilateral retinoblastoma.

Imaging included axial unenhanced T1-weighted images, T2-weighted images, postcontrast T1-weighted images, DWI, and ADC maps. The ADC maps were derived from the DWI automatically on a pixel-by-pixel basis.

The median ADC for trilateral retinoblastoma was 620.95x10-6 mm2/s with an interquartile range of 558.8x10-6 mm2/s, significantly lower than 1,238.5x10-6 mm2/s with an interquartile range of 433.35x10-6 mm2/s for normal pineal gland in bilateral retinoblastoma. (1)

Three Case Examples

These findings allowed us to confidently diagnose four of six patients in the trilateral retinoblastoma group. Two other patients, Patient A and Patient B, had equivocal imaging features. (1) However, we observed decreasing trends in diffusion restriction on serial MRI, which prompted consideration for trilateral retinoblastoma:

  • Patient A was a male infant diagnosed with bilateral retinoblastoma at 4 months old. Disease workup revealed no evidence of extraocular disease. An MRI brain scan found a pineal region focus of soft tissue measuring 0.7 x 0.5 centimeters at initial diagnosis that was not concerning. After genetic testing revealed a 13q14 small deletion, the patient received six cycles of systemic chemotherapy, three cycles of intra-arterial chemotherapy to the left eye, and multiple laser treatments to both eyes, which resolved the intraocular disease. At 14 months post diagnosis, MRI revealed a gradual increase in the size of the pineal mass. A repeat MRI at 17 months post diagnosis showed that it had not grown further but had a decreased ADC signal. A biopsy confirmed trilateral retinoblastoma. The patient successfully completed intensive chemotherapy according to the Children’s Oncology Group phase III protocol ARET0321 (NCT00554788) (Patient A was not enrolled in the study). At the 12-month follow-up MRI and ophthalmology exam, he was disease free. (1)
  • Patient B was a female infant diagnosed with bilateral retinoblastoma at 4 months old. She was treated successfully with four cycles of intra-arterial chemotherapy, laser therapy to both eyes, and cryotherapy to the left eye. The patient had yearly brain MRI scans for trilateral retinoblastoma surveillance. At month 27 postdiagnosis, imaging showed an increase in the size of the pineal gland, with portions of the gland showing increasingly restricted diffusion. Resection and pathology confirmed trilateral retinoblastoma. MRI spine and lumbar punctures were negative for the disease. The patient had intensive chemotherapy according to the same protocol as Patient A (Patient B was not enrolled in the study either). Unfortunately, 12 months later, she developed leptomeningeal recurrence and died from metastatic disease, despite undergoing whole-brain radiation and craniospinal radiation. (1)

In the group of eight patients with confirmed bilateral retinoblastoma, ADC mapping accurately diagnosed seven patients. (1) It also clarified equivocal results for a pineal mass for Patient C:

  • Patient C was closely followed with ophthalmology exams from birth because her father is a survivor of unilateral retinoblastoma. The patient was diagnosed with bilateral familial retinoblastoma at 1 month old, with no metastases evident on workup. The patient received six cycles of systemic chemotherapy, transpupillary thermotherapy, cryotherapy, and subconjunctival carboplatin injections. MRI brain scans at 7 months old showed a solid mass in the pineal region. Repeat MRI a month later showed no change in size, but the mass had developed cystic characteristics. The high ADC value remained unchanged, indicating that the pineal cyst was benign and the child would not need a biopsy or surgical resection. The patient was disease free at the one-year follow-up. (1)

Increasing Diagnostic Accuracy in the Future

The major limitation of our study is the small number of patients, but that was unavoidable given that trilateral retinoblastoma is such a rare cancer. However, our study includes the largest cohort of trilateral retinoblastoma patients to date.

Our investigation and results suggest that diffusion restriction and ADC trends can predict histologic change and subsequent malignant growth in intracranial pineal lesions that should require prompt consideration of trilateral retinoblastoma. Further, these advanced imaging techniques can distinguish benign pineal lesions that do not require invasive neurosurgical procedures.

We hope to see a prospective, multicenter study in the future to validate the diagnostic accuracy of ADC trends for predicting the risk of trilateral retinoblastoma. At MSK Kids, we are dedicated to finding a cure for children with retinoblastoma. Our Retinoblastoma Program, established more than 100 years ago in 1914, is the oldest program anywhere for the treatment of this rare pediatric disorder. Our goal is to help more pediatric patients achieve the best possible outcomes.

Refer a Patient
Call our dedicated clinician access number at 646-677-7440 or click the link below, and one of our care advisors will assist you with your referral needs.

The study authors declare no conflict of interest.

  1. Farouk SS, Haque S, Karimi S, et al. A potential role for apparent diffusion coefficient in the diagnosis of trilateral retinoblastoma. J Pediatr Hematol Oncol. June 27, 2019. (Epub ahead of print.) 
  2. de Jong MC, Kors WA, de Graaf P, et al. Trilateral retinoblastoma: a systematic review and meta-analysis. Lancet Oncol. 2014;15:1157–1167.
  3. Holladay DA, Holladay A, Montebello JF, et al. Clinical presentation, treatment, and outcome of trilateral retinoblastoma. Cancer. 1991;67:710–715.
  4. Amoaku WM, Willshaw HE, Parkes SE, et al. Trilateral retinoblastoma. A report of five patients. Cancer. 1996;78:858–863.
  5. Paulino AC. Trilateral retinoblastoma: is the location of the intracranial tumor important? Cancer. 1999;86:135–141.
  6. Blach LE, McCormick B, Abramson DH, et al. Trilateral retinoblastoma — incidence and outcome: a decade of experience. Int J Radiat Oncol Biol Phys. 1994;29:729–733.
  7. Kivela T. Trilateral retinoblastoma: a meta-analysis of hereditary retinoblastoma associated with primary ectopic intracranial retinoblastoma. J Clin Oncol. 1999;17:1829–1837.
  8. Dunkel IJ, Jubran RF, Gururangan S, et al. Trilateral retinoblastoma: potentially curable with intensive chemotherapy. Pediatr Blood Cancer. 2010;54:384–387.
  9. Wright KD, Qaddoumi I, Patay Z, et al. Successful treatment of early detected trilateral retinoblastoma using standard infant brain tumor therapy. Pediatr Blood Cancer. 2010;55:570–572.
  10. Rodjan F, de Graaf P, Brisse HJ, et al. Trilateral retinoblastoma: neuroimaging characteristics and value of routine brain screening on admission. J Neurooncol. 2012;109:535–544.
  11. Ramasubramanian A, Kytasty C, Meadows AT, et al. Incidence of pineal gland cyst and pineoblastoma in children with retinoblastoma during the chemoreduction era. Am J Ophthalmol. 2013;156:825–829.
  12. Beck Popovic M, Balmer A, Maeder P, et al. Benign pineal cysts in children with bilateral retinoblastoma: a new variant of trilateral retinoblastoma? Pediatr Blood Cancer. 2006;46: 755–761.
  13. Erdem E, Zimmerman RA, Haselgrove JC, et al. Diffusion-weighted imaging and fluid attenuated inversion recovery imaging in the evaluation of primitive neuroectodermal tumors. Neuroradiology. 2001;43:927–933.