Harvard Professor’s Papers Contain Copied Images, Says Science Sleuth

• doctor (v.)
• tranche of papers
• sleuth
• scrutinize
• malignant tumor
• glioblastomas
• sheer number of examples
• warrant (v.)
• anomaly
• human bone marrow
• carcinoma
•  

 

 

Harvard Professor’s Papers Contain Copied Images, Says Science Sleuth

Khalid Shah’s work is the latest tranche of papers to come under review

 

A Harvard Medical School scientist who studies deadly brain tumors is facing accusations that more than two dozen papers he co-authored contain scientific images that appear doctored or copied. 

Khalid Shah is vice chair for research at the Department of Neurosurgery at Brigham and Women’s Hospital and a professor at Harvard Medical School. Elisabeth Bik, a microbiologist and science image expert, this week sent accusations about 28 studies Shah co-authored to research integrity officials at both institutions and the journals that published them. 

Shah’s work is the latest tranche of papers to come under review by sleuths who are scrutinizing the scientific record for errors or fabricated data. In cases like Shah’s, the former scientists are taking advantage of advanced image analysis tools to spot copied images or scientific images that appear manipulated.  

Shah declined to comment on the accusations. Dr. Paul Anderson, chief academic officer at Mass General Brigham, the health system to which Brigham and Women’s belongs, said any research integrity allegations are addressed in a “robust and confidential process,” but didn’t comment on specifics about Shah’s case. 

A Harvard Medical School spokesperson said that “critical review of scientific work by peers—both before and after publication—is essential to the scientific process.”

Harvard Medical School PHOTO: BRIAN SNYDER/REUTERS

Shah hunts for ways to treat malignant brain tumors called glioblastomas using stem cells and gene editing tools. According to his website, he is a co-author on more than 100 papers in the fields of molecular biology and cancer, and holds more than 10 patents.

Shah’s work on stem cells to target tumors is the basis for the startup Amasa Therapeutics, according to the company’s website, where Shah is a member of the board of directors and has disclosed owning equity. 

Bik has been highlighting data irregularities in studies for more than a decade and has reported about 3,000 studies so far to journals or institutions. 

ADVERTISEMENT

 

Bik used ImageTwin and her own practiced eyes to examine 88 papers that included Shah as a co-author. The software compared images in studies against a database of 51 million scientific images dating back 20 years. Bik said she found signs of image manipulation, duplications or errors in 28 papers ranging from 2001 to 2023. 

Shah is listed as a first author or corresponding author on a majority of those 28 studies, which suggests he either was involved in the experiments or bore final responsibility for a study.

“Most of the problems in this set of 28 could be explained by honest error,” Bik said, for example if a researcher mislabeled their images and pasted in the wrong ones. She added that looking at published data alone makes it difficult to distinguish an error from misconduct. “There are a couple of papers that stand out that suggest an intention to mislead.”

“The sheer number of examples justify some concern,” said Matthew Schrag, a neurologist and researcher at the Vanderbilt University Medical Center, who outside his work at the institution reviewed Bik’s assessment. Schrag said he agreed with Bik’s observations in almost all cases and believed the issues warranted an institutional review. 

Bik on Tuesday posted her observations on PubPeer, an online forum that scientists use to question details in published studies. On Wednesday, she emailed her allegations to Harvard Medical School’s Office for Academic and Research Integrity and Mass General Brigham’s Anderson, and several journals.

One anomaly in the group is a 2022 paper in the journal Nature Communications which has images similar to those in nearly a dozen other sources, including papers published earlier, according to Bik. “I’ve never seen this,” she said.

In her PubPeer post, Elisabeth Bik highlighted in red one image published in a 2010 PLOS ONE paper that appears identical to an image published in a 2022 Nature Communications paper Khalid Shah co-authored.

 

Two images in the study matched photos on the website of ScienCell Research Laboratories, a Carlsbad, Calif., company that sells biomedical research supplies. Jennifer Welser, ScienCell’s vice president of research and development and scientific affairs, confirmed that the images on the website were made in-house by the company. The images show stem cells derived from human bone marrow which have distinctive features and make unique patterns within a dish, so she believed the images in Shah’s study were copied, she said. “We did not provide permission for these images to be used in this publication,” she said.

Chris Graf, research integrity director at Springer Nature, which publishes Nature Communications and some 3,000 other journals, said the company has received Bik’s complaint and is examining the studies she flagged.  

ADVERTISEMENT

 

Another image in the 2022 study, described as a microscope photograph of glioblastoma tissue, appeared similar to an image from a PLOS One study published in 2010. The older paper showed tissue from head and neck carcinomas, a different type of cancer.

Shi-Yong Sun, a cancer researcher at the Emory University School of Medicine and corresponding author on the PLOS One study, said he wasn’t aware of the image issue and didn’t know the Shah lab.  

Jennifer Byrne, a cancer researcher and research integrity adviser at the University of Sydney who reviewed this image set, agreed the images looked similar. “I can’t think of any reason they should be the same,” said Byrne, who helped develop software that screens papers for faulty DNA sequences.

Elisabeth Bik at her home in California. PHOTO: CLARA MOKRI FOR THE WALL STREET JOURNAL

In another instance, Bik highlighted similar images in two papers that shared some authors, including Shah. An image in a 2017 Scientific Reports paper describing an experiment with colorectal cancer cells looked similar to one published a few weeks later in a Neuro-Oncology study, involving different treatment conditions and glioblastoma cells.

Dr. Susan Chang, editor in chief of Neuro-Oncology, said the journal was made aware of the issue on Tuesday and plans to investigate. 

Bik said her analysis couldn’t tell who may be responsible for any errors in the studies. “That is up to the institution to investigate,” she said. 

Harvard Medical School has about 10,440 full-time campus and affiliate faculty, according to the school’s website. Misconduct accusations are first assessed for credibility by affiliate institutions where the work was conducted, and follow-up inquiries or investigations are managed by the medical school’s research integrity office, the medical school spokesperson said.

In January, Dana-Farber Cancer Institute said that 37 corrections or retraction requests were part of a review of more than 50 studies by four top doctors, also faculty at Harvard Medical School. Harvard Business School professor Francesca Gino was last year placed on administrative leave after data bloggers accused her work of containing falsified data. Gino has denied wrongdoing.

Write to Nidhi Subbaraman at nidhi.subbaraman@wsj.com